In the AIGC era, generating high-quality 4D content has garnered increasing research attention. Unfortunately, current 4D synthesis research is severely constrained by the lack of large-scale 4D datasets, preventing models from adequately learning the critical spatial-temporal features necessary for high-quality 4D generation, thus hindering progress in this domain. To combat this, we propose a novel framework that transfers rich spatial priors from existing 3D diffusion models and temporal priors from video diffusion models to enhance 4D synthesis. We develop a spatial-temporal-disentangled 4D (STD-4D) Diffusion model, which synthesizes 4D-aware videos through disentangled spatial and temporal latents. To facilitate the best feature transfer, we design a novel Orthogonal Spatial-temporal Distributional Transfer (Orster) mechanism, where the spatiotemporal feature distributions are carefully modeled and injected into the STD-4D Diffusion. Further, during the 4D construction, we devise a spatial-temporal-aware HexPlane (ST-HexPlane) to integrate the transferred spatiotemporal features for better 4D deformation and 4D Gaussian feature modeling. Experiments demonstrate that our method significantly outperforms existing approaches, achieving superior spatial-temporal consistency and higher-quality 4D synthesis.

Drag-Based Image Editing (DBIE), which allows users to manipulate images by directly dragging objects within them, has recently attracted much attention from the community. However, it faces two key challenges: (i) point-based drag is often highly ambiguous and difficult to align with user intentions; (ii) current DBIE methods primarily rely on alternating between motion supervision and point tracking, which is not only cumbersome but also fails to produce high-quality results. These limitations motivate us to explore DBIE from a new perspective---unifying it as a Latent Region Optimization (LRO) problem that aims to use region-level geometric transformations to optimize latent code to realize drag manipulation. Thus, by specifying the areas and types of geometric transformations, we can effectively address the ambiguity issue. We also propose a simple yet effective editing framework, dubbed DragNeXt. It solves LRO through Progressive Backward Self-Intervention (PBSI), simplifying the overall procedure of the alternating workflow while further enhancing quality by fully leveraging region-level structure information and progressive guidance from intermediate drag states. We validate DragNeXt on our NextBench, and extensive experiments demonstrate that our proposed method can significantly outperform existing approaches.

Bobo Li, Wu Rui, Zibo Ji, Meishan Zhang, Hao Fei, Min Zhang, Mong-Li Lee, Wynne Hsu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Li Zheng, Yanyi Luo, Hao Fei, Yuzhe Ding, Yujie Huang, Fei Li, Chong Teng, Donghong Ji. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Multimodal information extraction (IE) tasks have attracted increasing attention because many studies have shown that multimodal information benefits text information extraction. However, existing multimodal IE datasets mainly focus on sentence-level image-facilitated IE in English text, and pay little attention to video-based multimodal IE and fine-grained visual grounding. Therefore, in order to promote the development of multimodal IE, we constructed a multimodal multilingual multitask dataset, named M$^{3}$3D, which has the following features: (1) It contains paired document-level text and video to enrich multimodal information; (2) It supports two widely-used languages, namely English and Chinese; (3) It includes more multimodal IE tasks such as entity recognition, entity chain extraction, relation extraction and visual grounding. In addition, our dataset introduces an unexplored theme, i.e., biography, enriching the domains of multimodal IE resources. To establish a benchmark for our dataset, we propose an innovative hierarchical multimodal IE model. This model effectively leverages and integrates multimodal information through a Denoised Feature Fusion Module (DFFM). Furthermore, in non-ideal scenarios, modal information is often incomplete. Thus, we designed a Missing Modality Construction Module (MMCM) to alleviate the issues caused by missing modalities. Our model achieved an average performance of 53.80% and 53.77% on four tasks in English and Chinese datasets, respectively, which set a reasonable standard for subsequent research. In addition, we conducted more analytical experiments to verify the effectiveness of our proposed module. We believe that our work can promote the development of the field of multimodal IE.

Abstract Large Language Models (LLMs) exhibit unique strengths arising from differences in model architecture, training data, and strategies. Ensemble learning has been explored to leverage these complementary strengths through decision-level sharing (i.e.,Decision Ensemble), which combines the predictions from multiple LLMs. However, such methods integrate only shallow decisions and overlook the exchange of deeper levels of information within the internal representations of LLMs, such as problem understanding, world knowledge, and latent reasoning patterns. In this work, we propose Representation Ensemble (RISE), a novel ensemble framework that enables cross-LLM representation sharing for richer information exchange. To address challenges of representation-level interaction caused by layer misalignment and latent-space incompatibility across LLMs, we introduce a representation alignment method based on relational similarity measures and an orthogonal latent-space transformation. Experimental results show that (1) RISE achieves performance competitive with existing decision ensemble methods, and (2) RISE is strongly complementary to decision ensemble, with their combination boosting collaboration gains by 14%–41%. Finally, we further compare ensemble of small LLMs to a single larger LLM and to model merging and composition approaches, and find that ensemble learning consistently generalizes well without additional training.

In the realm of video dialog response generation, capturing both the essence of video content and the temporal nuances of conversation history is crucial. While some approaches rely on large-scale pretrained visual-language models, often neglecting temporal dynamics, others emphasize spatial-temporal relationships within videos but demand intricate object trajectory pre-extractions and overlook dialog temporal dynamics. This paper introduces the Dual Temporal Grounding-enhanced Video Dialog model (DTGVD), designed to bridge the gap between these two approaches. DTGVD uniquely integrates the strengths of both by emphasizing dual temporal relationships. It achieves this by predicting dialog turn-specific temporal regions, selectively filtering video content, and grounding responses in both video and dialog contexts.A key innovation of DTGVD is its advanced handling of chronological interplay within dialogs. By effectively capturing and leveraging dependencies between dialog turns, it enables a more nuanced understanding of conversational dynamics. To further align video and dialog temporal dynamics, we introduce a list-wise contrastive learning strategy. In this framework, accurately grounded turn-clip pairings are treated as positive samples, while less precise pairings serve as negative samples. This refined classification is then seamlessly integrated into our end-to-end response generation mechanism.Evaluations using AVSD@DSTC-7 and AVSD@DSTC-8 datasets underscore the superiority of our methodology.

Social Intelligence Queries (Social-IQ) serve as the primary multimodal benchmark for evaluating a model’s social intelligence level. While impressive multiple-choice question (MCQ) accuracy is achieved by current solutions, increasing evidence shows that they are largely, and in some cases entirely, dependent on language modality, overlooking visual context. Additionally, the closed-set nature further prevents the exploration of whether and to what extent the reasoning path behind selection is correct. To address these limitations, we propose the Visually Explainable and Grounded Artificial Social Intelligence (VEGAS) model. As a generative multimodal model, VEGAS leverages open-ended answering to provide explainable responses, which enhances the clarity and evaluation of reasoning paths. To enable visually grounded answering, we propose a novel sampling strategy to provide the model with more relevant visual frames. We then enhance the model’s interpretation of these frames through Generalist Instruction Fine-Tuning (GIFT), which aims to: i) learn multimodal language transformations for fundamental emotional social traits, and ii) establish multimodal joint reasoning capabilities. Extensive experiments, comprising modality ablation, open-ended assessments, and supervised MCQ evaluations, consistently show that VEGAS effectively utilizes visual information in reasoning to produce correct and also credible answers. We expect this work to offer a new perspective on Social-IQ and advance the development of human-like social AI.

Recent advancements in multimodal large language models (MLLMs) have shown unprecedented capabilities in advancing various vision-language tasks. However, MLLMs face significant challenges with hallucinations, and misleading outputs that do not align with the input data. While existing efforts are paid to combat MLLM hallucinations, several pivotal challenges are still unsolved. First, while current approaches aggressively focus on addressing errors at the perception level, another important type at the cognition level requiring factual commonsense can be overlooked. In addition, existing methods might fall short in finding a more effective way to represent visual input, which is yet a key bottleneck that triggers visual hallucinations. Moreover, MLLMs can frequently be misled by faulty textual inputs and cause hallucinations, while unfortunately, this type of issue has long been overlooked by existing studies. Inspired by human intuition in handling hallucinations, this paper introduces a novel bottom-up reasoning framework. Our framework systematically addresses potential issues in both visual and textual inputs by verifying and integrating perception-level information with cognition-level commonsense knowledge, ensuring more reliable outputs. Extensive experiments demonstrate significant improvements in multiple hallucination benchmarks after integrating MLLMs with the proposed framework. In-depth analyses reveal the great potential of our methods in addressing perception- and cognition-level hallucinations.

With the continuous emergence of various social media platforms frequently used in daily life, the multimodal meme understanding (MMU) task has been garnering increasing attention. MMU aims to explore and comprehend the meanings of memes from various perspectives by performing tasks such as metaphor recognition, sentiment analysis, intention detection, and offensiveness detection. Despite making progress, limitations persist due to the loss of fine-grained metaphorical visual clue and the neglect of multimodal text-image weak correlation. To overcome these limitations, we propose a multi-granular multimodal clue fusion model (MGMCF) to advance MMU. Firstly, we design an object-level semantic mining module to extract object-level image feature clues, achieving fine-grained feature clue extraction and enhancing the model's ability to capture metaphorical details and semantics. Secondly, we propose a brand-new global-local cross-modal interaction model to address the weak correlation between text and images. This model facilitates effective interaction between global multimodal contextual clues and local unimodal feature clues, strengthening their representations through a bidirectional cross-modal attention mechanism. Finally, we devise a dual-semantic guided training strategy to enhance the model's understanding and alignment of multimodal representations in the semantic space. Experiments conducted on the widely-used MET-MEME bilingual dataset demonstrate significant improvements over state-of-the-art baselines. Specifically, there is an 8.14% increase in precision for offensiveness detection task, and respective accuracy enhancements of 3.53%, 3.89%, and 3.52% for metaphor recognition, sentiment analysis, and intention detection tasks. These results, underpinned by in-depth analyses, underscore the effectiveness and potential of our approach for advancing MMU.

Jundong Xu, Hao Fei, Meng Luo, Qian Liu, Liangming Pan, William Yang Wang, Preslav Nakov, Mong-Li Lee, Wynne Hsu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Li Zheng, Sihang Wang, Hao Fei, Zuquan Peng, Fei Li, Jianming Fu, Chong Teng, Donghong Ji. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Scene graph (SG) representations can neatly and efficiently describe scene semantics, which has driven sustained intensive research in SG generation. In the real world, multiple modalities often coexist, with different types, such as images, text, video, and 3D data, expressing distinct characteristics. Unfortunately, current SG research is largely confined to single-modality scene modeling, preventing the full utilization of the complementary strengths of different modality SG representations in depicting holistic scene semantics. To this end, we introduce Universal SG (USG), a novel representation capable of fully characterizing comprehensive semantic scenes from any given combination of modality inputs, encompassing modality-invariant and modality-specific scenes. Further, we tailor a niche-targeting USG parser, USG-Par, which effectively addresses two key bottlenecks of cross-modal object alignment and out-of-domain challenges. We design the USG-Par with modular architecture for end-to-end USG generation, in which we devise an object associator to relieve the modality gap for cross-modal object alignment. Further, we propose a text-centric scene contrasting learning mechanism to mitigate domain imbalances by aligning multimodal objects and relations with textual SGs. Through extensive experiments, we demonstrate that USG offers a stronger capability for expressing scene semantics than standalone SGs, and also that our USG-Par achieves higher efficacy and performance. The project page is https://sqwu.top/USG/.

The latest emerged 4D Panoptic Scene Graph (4D-PSG) provides an advanced-ever representation for comprehensively modeling the dynamic 4D visual real world. Unfortunately, current pioneering 4D-PSG research can primarily suffer from data scarcity issues severely, as well as the resulting out-of-vocabulary problems; also, the pipeline nature of the benchmark generation method can lead to suboptimal performance. To address these challenges, this paper investigates a novel framework for 4D-PSG generation that leverages rich 2D visual scene annotations to enhance 4D scene learning. First, we introduce a 4D Large Language Model (4D-LLM) integrated with a 3D mask decoder for end-to-end generation of 4D-PSG. A chained SG inference mechanism is further designed to exploit LLMs’ open-vocabulary capabilities to infer accurate and comprehensive object and relation labels iteratively. Most importantly, we propose a 2D-to-4D visual scene transfer learning framework, where a spatial-temporal scene transcending strategy effectively transfers dimension-invariant features from abundant 2D SG annotations to 4D scenes, effectively compensating for data scarcity in 4D-PSG. Extensive experiments on the benchmark data demonstrate that we strikingly outperform baseline models by a large margin, highlighting the effectiveness of our method. The project page is https://sqwu.top/PSG-4D-LLM/.

Enabling digital humans to express rich emotions has significant applications in dialogue systems, gaming, and other interactive scenarios.While recent advances in talking head synthesis have achieved impressive results in lip synchronization, they tend to overlook the rich and dynamic nature of facial expressions.To fill this critical gap, we introduce an endto-end text-to-expression model that explicitly focuses on emotional dynamics.Our model learns expressive facial variations in a continuous latent space and generates expressions that are diverse, fluid, and emotionally coherent.To support this task, we introduce EmoAva, a large-scale and high-quality dataset containing 15,000 text-3D expression pairs.Extensive experiments on both existing datasets and EmoAva demonstrate that our method significantly outperforms baselines across multiple evaluation metrics, marking a significant advancement in the field.

Open Information Extraction (OIE) aims at extracting the relational triplets from open-domain texts. Existing methods, unfortunately, mostly fall prey to the complex OIE setting, due to the failure to extract unseen words and underutilize syntactic features. In this work, we propose a novel system tailored for complex OIE, where a generative PLM with non-autoregressive generative decoding is adopted for abstractive OIE generation. We introduce a heterogeneous syntactic forests as features for the task, merging constituency and dependency forests into a unified syntax graph, aiding in better detecting potential boundaries and relationships within complex terms. Also an Implicit-Explicit Contrastive Learning mechanism is devised to boost the perception of implicit relations and spans. Our system outperforms the current state-of-the-art model across seven OIE datasets.

Digital agents are increasingly employed to automate tasks in interactive digital environments such as web pages, software applications, and operating systems. While text-based agents built on Large Language Models (LLMs) often require frequent updates due to platform-specific APIs, visual agents leveraging Multimodal Large Language Models (MLLMs) offer enhanced adaptability by interacting directly with Graphical User Interfaces (GUIs). However, these agents face significant challenges in visual perception, particularly when handling high-resolution, visually complex digital environments. This paper introduces Iris, a foundational visual agent that addresses these challenges through two key innovations: Information-Sensitive Cropping (ISC) and Self-Refining Dual Learning (SRDL). ISC dynamically identifies and prioritizes visually dense regions using a edge detection algorithm, enabling efficient processing by allocating more computational resources to areas with higher information density. SRDL enhances the agent's ability to handle complex tasks by leveraging a dual-learning loop, where improvements in referring (describing UI elements) reinforce grounding (locating elements) and vice versa, all without requiring additional annotated data. Empirical evaluations demonstrate that Iris achieves state-of-the-art performance across multiple benchmarks with only 850K GUI annotations, outperforming methods using 10x more training data. These improvements further translate to significant gains in both web and OS agent downstream tasks.

The emergence of vision-language models has transformed medical AI, enabling unprecedented advances in diagnostic capability and clinical applications. However, progress in dermatology has lagged behind other medical domains due to the lack of standard image-text pairs. Existing dermatological datasets are limited in both scale and depth, offering only single-label annotations across a narrow range of diseases instead of rich textual descriptions, and lacking the crucial clinical context needed for real-world applications. To address these limitations, we present Derm1M, the first large-scale vision-language dataset for dermatology, comprising 1,029,761 image-text pairs. Built from diverse educational resources and structured around a standard ontology collaboratively developed by experts, Derm1M provides comprehensive coverage for over 390 skin conditions across four hierarchical levels and 130 clinical concepts with rich contextual information such as medical history, symptoms, and skin tone. To demonstrate Derm1M potential in advancing both AI research and clinical application, we pretrained a series of CLIP-like models, collectively called DermLIP, on this dataset. The DermLIP family significantly outperforms state-of-the-art foundation models on eight diverse datasets across multiple tasks, including zero-shot skin disease classification, clinical and artifacts concept identification, few-shot/full-shot learning, and cross-modal retrieval. Our dataset and code will be publicly available at https://github.com/SiyuanYan1/Derm1M upon acceptance.

Modeling task-driven attention in driving is a fundamental challenge for both autonomous vehicles and cognitive science. Existing methods primarily predict where drivers look by generating spatial heatmaps, but fail to capture the cognitive motivations behind attention allocation in specific contexts, which limits deeper understanding of attention mechanisms. To bridge this gap, we introduce Explainable Driver Attention Prediction, a novel task paradigm that jointly predicts spatial attention regions (where), parses attended semantics (what), and provides cognitive reasoning for attention allocation (why). To support this, we present W3DA, the first large-scale explainable driver attention dataset. It enriches existing benchmarks with detailed semantic and causal annotations across diverse driving scenarios, including normal conditions, safety-critical situations, and traffic accidents. We further propose LLada, a Large Language model-driven framework for driver attention prediction, which unifies pixel modeling, semantic parsing, and cognitive reasoning within an end-to-end architecture. Extensive experiments demonstrate the effectiveness of LLada, exhibiting robust generalization across datasets and driving conditions. This work serves as a key step toward a deeper understanding of driver attention mechanisms, with significant implications for autonomous driving, intelligent driver training, and human-computer interaction.

Multimodal Large Language Models (MLLMs) still struggle with hallucinations despite their impressive capabilities. Recent studies have attempted to mitigate this by applying Direct Preference Optimization (DPO) to multimodal scenarios using preference pairs from text-based responses. However, our analysis of representation distributions reveals that multimodal DPO struggles to align image and text representations and to distinguish between hallucinated and non-hallucinated descriptions. To address these challenges, In this work, we propose a Cross-modal Hierarchical Direct Preference Optimization (CHiP) to address these limitations. We introduce a visual preference optimization module within the DPO framework, enabling MLLMs to learn from both textual and visual preferences simultaneously. Furthermore, we propose a hierarchical textual preference optimization module that allows the model to capture preferences at multiple granular levels, including response, segment, and token levels. We evaluate CHiP through both quantitative and qualitative analyses, with results across multiple benchmarks demonstrating its effectiveness in reducing hallucinations. On the Object HalBench dataset, CHiP outperforms DPO in hallucination reduction, achieving improvements of 52.7% and 55.5% relative points based on the base model Muffin and LLaVA models, respectively. We make all our datasets and code publicly available.

Multimodal Large Language Models (MLLMs) have demonstrated exceptional capabilities in processing vision-language tasks. One of the crux of MLLMs lies in vision tokenization, which involves efficiently transforming input visual signals into feature representations that are most beneficial for LLMs. However, existing vision tokenizers, essential for semantic alignment between vision and language, remain problematic. Existing methods aggressively fragment visual input, corrupting the visual semantic integrity. To address this, this paper proposes a novel dynamic Semantic-Equivalent Vision Tokenizer (SeTok), which groups visual features into semantic units via a dynamic clustering algorithm, flexibly determining the number of tokens based on image complexity. The resulting vision tokens effectively preserve semantic integrity and capture both low-frequency and high-frequency visual features. The proposed MLLM (Setokim) equipped with SeTok significantly demonstrates superior performance across various tasks, as evidenced by our experimental results.

The Multimodal Large Language Model (MLLM) is currently experiencing rapid growth, driven by the advanced capabilities of language-based LLMs. Unlike their specialist predecessors, existing MLLMs are evolving towards a Multimodal Generalist paradigm. Initially limited to understanding multiple modalities, these models have advanced to not only comprehend but also generate across modalities. Their capabilities have expanded from coarse-grained to fine-grained multimodal understanding and from supporting singular modalities to accommodating a wide array of or even arbitrary modalities. To assess the capabilities of various MLLMs, a diverse array of benchmark test sets has been proposed. This leads to a critical question: Can we simply assume that higher performance across tasks indicates a stronger MLLM capability, bringing us closer to human-level AI? We argue that the answer is not as straightforward as it seems. In this project, we introduce an evaluation framework to delineate the capabilities and behaviors of current multimodal generalists. This framework, named General-Level, establishes 5-scale levels of MLLM performance and generality, offering a methodology to compare MLLMs and gauge the progress of existing systems towards more robust multimodal generalists and, ultimately, towards AGI (Artificial General Intelligence). Central to our framework is the use of Synergy as the evaluative criterion, categorizing capabilities based on whether MLLMs preserve synergy across comprehension and generation, as well as across multimodal interactions. To evaluate the comprehensive abilities of various generalists, we present a massive multimodal benchmark, General-Bench, which encompasses a broader spectrum of skills, modalities, formats, and capabilities, including over 700 tasks and 325,800 instances. The evaluation results that involve over 100 existing state-of-the-art MLLMs uncover the capability rankings of generalists, highlighting the challenges in reaching genuine AI. We expect this project to pave the way for future research on next-generation multimodal foundation models, providing a robust infrastructure to accelerate the realization of AGI. Project Page: https://generalist.top/, Leaderboard: https://generalist.top/leaderboard/, Benchmark: https://huggingface.co/General-Level/.

Multi-Modal Large Language Models (MLLMs) have exhibited remarkable performance on various vision-language tasks such as Visual Question Answering (VQA). Despite accumulating evidence of privacy concerns associated with task-relevant content, it remains unclear whether MLLMs inadvertently memorize private content that is entirely irrelevant to the training tasks. In this paper, we investigate how randomly generated task-irrelevant private content can become spuriously correlated with downstream objectives due to partial mini-batch training dynamics, thus causing inadvertent memorization. Concretely, we randomly generate task-irrelevant watermarks into VQA fine-tuning images at varying probabilities and propose a novel probing framework to determine whether MLLMs have inadvertently encoded such content. Our experiments reveal that MLLMs exhibit notably different training behaviors in partial mini-batch settings with task-irrelevant watermarks embedded. Furthermore, through layer-wise probing, we demonstrate that MLLMs trigger distinct representational patterns when encountering previously seen task-irrelevant knowledge, even if this knowledge does not influence their output during prompting. Our code is available at https://github.com/illusionhi/ProbingPrivacy.

The utilization of automated detection of fake news videos significantly improves early intervention for misinformation on short video platforms. A shortcoming of existing approaches is their inability to fuse multimodal features effectively. They only perform simple fusion of multimodal features to predict fake news, without deeply considering inter-modality relations. Inspired by the way people read news videos, we propose a novel multimodal deep fusion network (MFSVFND) that can effectively capture the important details of unimodality in detecting fake news videos by learning inter-dependencies among multimodal features. We conducted extensive experiments on two large-scale dataset of fake news videos in different languages, and our model outperforms state-of-the-art methods, demonstrating the superiority of MFSVFND in detecting fake news on short video platforms.

The 1st Workshop on Cognition-oriented Multimodal Affective and Empathetic Computing (CogMAEC) was held at ACM Multimedia 2025. It focused on moving emotional AI beyond basic recognition toward deeper cognitive understanding. While traditional multimodal affective computing has emphasized simple emotion detection, the rise of multimodal large language models (MLLMs) has spurred interest in modeling how emotions emerge and evolve in context. The workshop gathered researchers on emotion reasoning, multimodal understanding, and human-computer empathy, exploring how machines can not only recognize emotions but also explain their causes and simulate human-like affective reasoning. The program featured invited talks, oral presentations, and posters spanning perception, interaction, causal modeling, and cognitive grounding. CogMAEC provided a platform to connect researchers across disciplines and foster future work on cognitively aware affective computing. Materials are available at https://CogMAEC.github.io/MM2025.

Online form filling is a common yet labor-intensive task involving extensive keyboard and mouse interactions. Despite the long-standing vision of automating this process with ''one click,'' existing tools remain largely rule-based and lack generalizable, generative capabilities. Recent advances in Multimodal Large Language Models (MLLMs) have enabled promising agents for GUI-related tasks in general-purpose scenarios. However, they struggle with the unique challenges of form filling, such as flexible layouts and the difficulty of aligning textual instructions with on-screen fields. To bridge this gap, we formally define the form-filling task and propose FormFactory-an interactive benchmarking suite comprising a web-based interface, backend evaluation module, and carefully constructed dataset. Our benchmark covers diverse real-world scenarios, incorporates various field formats, and simulates high-fidelity form interactions. We conduct a comprehensive evaluation of state-of-the-art MLLMs and observe that no model surpasses 5% accuracy, underscoring the inherent difficulty of the task. These findings also reveal significant limitations in current models' visual layout reasoning and field-value alignment abilities.

Understanding fine-grained sentiment dynamics in human conversations is a central goal for next-generation artificial intelligence, especially in scenarios where interactions are rich in both modalities and context. To advance research in this area, we organize the Multimodal Conversational Aspect-based Sentiment Analysis (MCABSA) challenge to the community of aspect-based sentiment analysis. The MCABSA challenge introduces two novel subtasks: 1) Panoptic Sentiment Sextuple Extraction, panoramically recognizing holder, target, aspect, opinion, sentiment, and rationale from multi-turn, multi-party multimodal dialogue; and 2) Sentiment Flipping Analysis, detecting the dynamic sentiment transformation throughout the conversation along with the causal reasons. To support these tasks, we present the PanoSent dataset, a high-quality, large-scale benchmark featuring multi-turn, multi-party dialogues annotated with both explicit and implicit sentiment elements across text, image, audio, and video modalities. PanoSent offers extensive real-world scenario coverage, providing a comprehensive testbed for multimodal conversational sentiment analysis. The challenge has attracted widespread participation from both academia and industry, with over 30 teams registered and more than 100 successful submissions. In this paper, we introduce the task, dataset, and evaluation settings, summarize the systems of the top teams, and discuss the findings of the participants. Further details of the challenge can be found at https://panosent.github.io/MM25-challenge.

RGB-D salient object detection (SOD) aims to identify the most conspicuous objects in a scene with the incorporation of depth cues. Existing methods mainly rely on CNNs, limited by the local receptive fields, or Vision Transformers that suffer from the cost of quadratic complexity, posing a challenge in balancing performance and computational efficiency. Recently, state space models (SSM), Mamba, have shown great potential for modeling long-range dependency with linear complexity. However, directly applying SSM to RGB-D SOD may lead to deficient local semantics as well as the inadequate cross-modality fusion. To address these issues, we propose a Local Emphatic and Adaptive Fusion state space model (LEAF-Mamba) that contains two novel components: 1) a local emphatic state space module (LE-SSM) to capture multi-scale local dependencies for both modalities. 2) an SSM-based adaptive fusion module (AFM) for complementary cross-modality interaction and reliable cross-modality integration. Extensive experiments demonstrate that the LEAF-Mamba consistently outperforms 16 state-of-the-art RGB-D SOD methods in both efficacy and efficiency. Moreover, our method can achieve excellent performance on the RGB-T SOD task, proving a powerful generalization ability. Our code is publicly available at https://github.com/LanhooNg/LEAF-Mamba.

The Ava-MERG Challenge at ACM Multimedia 2025 explores avatar-based multimodal empathetic response generation by advancing dialogue systems that respond with emotional awareness across text, speech, and talking-face video. To support this goal, we constructed a high-quality dataset featuring aligned multimodal responses in diverse conversational scenarios. The challenge comprises two progressively complex subtasks centering around multimodal-aware empathetic response generation. We aim to promote research in empathetic AI and affective computing.

Video understanding plays a vital role in bridging low-level visual signals with high-level cognitive reasoning, and is fundamental to applications such as autonomous driving, embodied AI, and the broader pursuit of AGI. The rapid development of large language models (LLMs), particularly those utilizing Chain-of-Thought (CoT) technology, has significantly advanced video reasoning capabilities. However, current approaches primarily depend on textual information for reasoning, overlooking the visual modality in the actual video reasoning process. In contrast, humans naturally re-examine visual content while reasoning. Motivated by this, we introduce a novel video reasoning paradigm: Video-Text Interleaved CoT (ViTCoT), which facilitates more intuitive and cognitively aligned reasoning. To the end, first, we construct the Video-Text Interleaved Benchmark (ViTIB), which is created using MLLMs for key-video selection and manually verified. Furthermore, we extensively explore the potential of the ViTCoT paradigm in the video understanding field. Extensive experiments demonstrate that ViTCoT significantly enhances performance compared to the traditional text-only CoT paradigm and effectively activates more neuron values in MLLMs.

This paper presents JavisGPT, the first unified multimodal large language model (MLLM) for joint audio-video (JAV) comprehension and generation. JavisGPT has a concise encoder-LLM-decoder architecture, which has a SyncFusion module for spatio-temporal audio-video fusion and synchrony-aware learnable queries to bridge a pretrained JAV-DiT generator. This design enables temporally coherent video-audio understanding and generation from multimodal instructions. We design an effective three-stage training pipeline consisting of multimodal pretraining, audio-video fine-tuning, and large-scale instruction-tuning, to progressively build multimodal comprehension and generation from existing vision-language models. For instruction tuning, we construct JavisInst-Omni, a high-quality instruction dataset with over 200K GPT-4o-curated audio-video-text dialogues that cover diverse and multi-level comprehension and generation scenarios. On JAV comprehension and generation benchmarks, our experiments show that JavisGPT outperforms existing MLLMs, particularly in complex and temporally synchronized settings.

Multimodal symbolic logical reasoning, which aims to deduce new facts from multimodal input via formal logic, is critical in high-stakes applications such as autonomous driving and medical diagnosis, as its rigorous, deterministic reasoning helps prevent serious consequences. To evaluate such capabilities of current state-of-the-art vision language models (VLMs), we introduce the first benchmark MuSLR for multimodal symbolic logical reasoning grounded in formal logical rules. MuSLR comprises 1,093 instances across 7 domains, including 35 atomic symbolic logic and 976 logical combinations, with reasoning depths ranging from 2 to 9. We evaluate 7 state-of-the-art VLMs on MuSLR and find that they all struggle with multimodal symbolic reasoning, with the best model, GPT-4.1, achieving only 46.8%. Thus, we propose LogiCAM, a modular framework that applies formal logical rules to multimodal inputs, boosting GPT-4.1’s Chain-of-Thought performance by 14.13%, and delivering even larger gains on complex logics such as first-order logic. We also conduct a comprehensive error analysis, showing that around 70% of failures stem from logical misalignment between modalities, offering key insights to guide future improvements.

This paper introduces **VimoRAG**, a novel video-based retrieval-augmented motion generation framework for motion large language models (LLMs). As motion LLMs face severe out-of-domain/out-of-vocabulary issues due to limited annotated data, **VimoRAG** leverages large-scale in-the-wild video databases to enhance 3D motion generation by retrieving relevant 2D human motion signals. While video-based motion RAG is nontrivial, we address two key bottlenecks: (1) developing an effective motion-centered video retrieval model that distinguishes human poses and actions, and (2) mitigating the issue of error propagation caused by suboptimal retrieval results. We design the Gemini Motion Video Retriever mechanism and the Motion-centric Dual-alignment DPO Trainer, enabling effective retrieval and generation processes. Experimental results show that **VimoRAG** significantly boosts the performance of motion LLMs constrained to text-only input.

In text-based person search endeavors, data generation has emerged as a prevailing practice, addressing concerns over privacy preservation and the arduous task of manual annotation. Although the number of synthesized data can be infinite in theory, the scientific conundrum persists that how much generated data optimally fuels subsequent model training. We observe that only a subset of the data in these constructed datasets plays a decisive role. Therefore, we introduce a new Filtering-WoRA paradigm, which contains a filtering algorithm to identify this crucial data subset and WoRA (Weighted Low-Rank Adaptation) learning strategy for light fine-tuning. The filtering algorithm is based on the cross-modality relevance to remove the lots of coarse matching synthesis pairs. As the number of data decreases, we do not need to fine-tune the entire model. Therefore, we propose a WoRA learning strategy to efficiently update a minimal portion of model parameters. WoRA streamlines the learning process, enabling heightened efficiency in extracting knowledge from fewer, yet potent, data instances. Extensive experimentation validates the efficacy of pretraining, where our model achieves advanced and efficient retrieval performance on challenging real-world benchmarks. Notably, on the CUHK-PEDES dataset, we have achieved a competitive mAP of 67.02% while reducing model training time by 19.82%.

Empathetic Response Generation (ERG) is one of the key tasks of the affective computing area, which aims to produce emotionally nuanced and compassionate responses to user's queries. However, existing ERG research is predominantly confined to the singleton text modality, limiting its effectiveness since human emotions are inherently conveyed through multiple modalities. To combat this, we introduce an avatar-based Multimodal ERG (MERG) task, entailing rich text, speech, and facial vision information. We first present a large-scale high-quality benchmark dataset, AvaMERG, which extends traditional text ERG by incorporating authentic human speech audio and dynamic talking-face avatar videos, encompassing a diverse range of avatar profiles and broadly covering various topics of real-world scenarios. Further, we deliberately tailor a system, named Empatheia, for MERG. Built upon a Multimodal Large Language Model (MLLM) with multimodal encoder, speech and avatar generators, Empatheia performs end-to-end MERG, with Chain-of-Empathetic reasoning mechanism integrated for enhanced empathy understanding and reasoning.Finally, we devise a list of empathetic-enhanced tuning strategies, strengthening the capabilities of emotional accuracy and content, avatar-profile consistency across modalities. Experimental results on AvaMERG data demonstrate that Empatheia consistently shows superior performance than baseline methods on both textual ERG and MERG. All data and code are open at https://AvaMERG.github.io/.

Abstract Character Coreference Resolution in Movie Screenplays (MovieCoref) is a newly emerging task for understanding complex movie plots and character relationships. This task poses greater challenges than traditional coreference resolution, due to the intricate narrative structures and character interactions unique to screenplays. In light of these challenges, we introduce a novel approach: a Structure-aware Triaffine Parser (STPar) for the MovieCoref task. STPar combines discourse and syntactic structures in the feature encoding process, enabling comprehensive analysis of ternary relationships and complex interactions. During the pairing process, STPar utilizes a triaffine scorer to consider high-order relations between candidate mention pairs, thus enhancing its ability to capture detailed narrative structures. In addition, STPar incorporates multi-task learning, encompassing singleton and span detection tasks, to further improve coreference resolution performance. Our evaluations on the MovieCoref dataset demonstrate that STPar significantly outperforms the best baseline by 7.4%, 21.5%, 7.1%, and 10.2% in F1 scores of B3, CEAFe, LEA, and CoNLL. Further analysis highlights the benefits of integrating structural discourse and syntactic information as well as the combined approaches of triaffine and multi-task learning.1

Dialogue disentanglement aims to detach the chronologically ordered utterances into several independent sessions. Conversation utterances are essentially organized and described by the underlying discourse, and thus dialogue disentanglement requires the full understanding and harnessing of the intrinsic discourse attribute. In this article, we propose enhancing dialogue disentanglement by taking full advantage of the dialogue discourse characteristics. First of all, in feature encoding stage , we construct the heterogeneous graph representations to model the various dialogue-specific discourse structural features, including the static speaker-role structures (i.e., speaker-utterance and speaker-mentioning structure) and the dynamic contextual structures (i.e., the utterance-distance and partial-replying structure). We then develop a structure-aware framework to integrate the rich structural features for better modeling the conversational semantic context. Second, in model learning stage , we perform optimization with a hierarchical ranking loss mechanism, which groups dialogue utterances into different discourse levels and carries training covering pairwise and session-wise levels hierarchically. Third, in inference stage , we devise an easy-first decoding algorithm, which performs utterance pairing under the easy-to-hard manner with a global context, breaking the constraint of traditional sequential decoding order. On two benchmark datasets, our overall system achieves new state-of-the-art performances on all evaluations. In-depth analyses further demonstrate the efficacy of each proposed idea and also reveal how our methods help advance the task. Our work has great potential to facilitate broader multi-party multi-thread dialogue applications.

This editorial introduces the Special Issue on Deep Multimodal Generation and Retrieval, hosted by the ACM Transactions on Multimedia Computing, Communications, and Applications in 2024. Information generation (IG) and information retrieval (IR) are two key representative approaches of information acquisition, i.e., producing content either via generation or via retrieval. While traditional IG and IR have achieved great success within the scope of languages, the under-utilization of varied data sources in different modalities (i.e., text, images, audio, and video) would hinder IG and IR techniques from giving the full advances and thus limit the applications in the real world. Knowing the fact that our world is replete with multimedia information, this special issue encourages the development of deep multimodal learning for the research of IG and IR. Benefiting from a variety of data types and modalities, some of the latest prevailing techniques are extensively invented to show great facilitation in multimodal IG and IR learning. With this special issue, we encourage explorations in Deep Multimodal Generation and Retrieval, providing a platform for researchers to share insights and advancements in this rapidly evolving domain. Each article provides novel insights into areas and challenges such as Multimodal Semantics Understanding, Generative Models for Vision Synthesis, Multimodal Information Retrieval, Explainable and Reliable Multimodal Learning. We summarize the main contributions of the included works and emphasize their role in advancing the field of multimodal generation or via retrieval. Finally, we discuss ongoing challenges and future opportunities in this rapidly evolving domain, particularly in the context of large foundational models.

Referring Image Segmentation (RIS) has been extensively studied over the past decade, leading to the development of advanced algorithms. However, there has been a lack of research investigating how existing algorithms should be benchmarked with complex language queries, which include more informative descriptions of surrounding objects and backgrounds (e.g., the black car vs. the black car is parking on the road and beside the bus ). Given the significant improvement in the semantic understanding capability of large pre-trained models, it is crucial to take a step further in RIS by incorporating complex language that resembles real-world applications. To close this gap, building upon the existing RefCOCO and Visual Genome datasets, we propose a new RIS benchmark with complex queries, namely RIS-CQ . The RIS-CQ dataset is of high quality and large scale, which challenges the existing RIS with enriched, specific, and informative queries, and enables a more realistic scenario of RIS research. Besides, we present a niche targeting method to better task the RIS-CQ, called Dual-Modality Graph Alignment ( DuMoGa ) model, which outperforms a series of RIS methods. To provide a valuable foundation for future advancements in the field of RIS with complex queries, we release the datasets, pre-processing and synthetic scripts, and the algorithm implementations at https://github.com/lili0415/DuMoGa .

The multimodal emotion-cause pair extraction (MECPE) task aims to detect the emotions, causes, and emotion-cause pairs from multimodal conversations. Existing methods for this task typically concatenate representations of each utterance from distinct modalities and then predict emotion-cause pairs directly. This approach struggles to effectively integrate multimodal features and capture the subtleties of emotion transitions, which are crucial for accurately identifying causes—thereby limiting overall performance. To address these challenges, we propose a novel model that captures holistic interaction and label constraint (HiLo) features for the MECPE task. HiLo facilitates cross-modality and cross-utterance feature interaction with various attention mechanisms, establishing a robust foundation for precise cause extraction. Notably, our model innovatively leverages emotion transition features as pivotal cues to enhance causal inference within conversations. The experimental results demonstrate the superior performance of HiLo, evidenced by an increase of more than 2% in the F1 score compared to existing benchmarks. Further analysis reveals that our approach adeptly utilizes multimodal and dialogue features, making a significant contribution to the field of emotion-cause analysis. Our code is publicly available at https://is.gd/MVdYmx .

Multi-modal sarcasm detection involves determining whether a given multi-modal input conveys sarcastic intent by analyzing the underlying sentiment. Recently, vision large language models have shown remarkable success on various of multi-modal tasks. Inspired by this, we systematically investigate the impact of vision large language models in zero-shot multi-modal sarcasm detection task. Furthermore, to capture different perspectives of sarcastic expressions, we propose a multi-view agent framework, S 3 Agent, designed to enhance zero-shot multi-modal sarcasm detection by leveraging three critical perspectives: superficial expression , semantic information , and sentiment expression . Our experiments on the MMSD2.0 dataset, which involves six models and four prompting strategies, demonstrate that our approach achieves state-of-the-art performance. Our method achieves an average improvement of 13.2% in accuracy. Moreover, we evaluate our method on the text-only sarcasm detection task, where it also surpasses baseline approaches.

Dialogue Aspect-based Sentiment Quadruple (DiaASQ) is a newly-emergent task aiming to extract the sentiment quadruple (i.e., targets, aspects, opinions, and sentiments) from conversations. While showing promising performance, the prior DiaASQ approach unfortunately falls prey to the key crux of DiaASQ, including insufficient modeling of discourse features, and lacking quadruple extraction, which hinders further task improvement. To this end, we introduce a novel framework that not only capitalizes on comprehensive discourse feature modeling, but also captures the intrinsic interaction for optimal quadruple extraction. On the one hand, drawing upon multiple discourse features, our approach constructs a token-level heterogeneous graph and enhances token interactions through a heterogeneous attention network. We further propose a novel triadic scorer, strengthening weak token relations within a quadruple, thereby enhancing the cohesion of the quadruple extraction. Experimental results on the DiaASQ benchmark showcase that our model significantly outperforms existing baselines across both English and Chinese datasets. Our code is available at https://bit.ly/3v27pqA.

In recent years, spectral graph neural networks, characterized by polynomial filters, have garnered increasing attention and have achieved remarkable performance in tasks such as node classification. These models typically assume that eigenvalues for the normalized Laplacian matrix are distinct from each other, thus expecting a polynomial filter to have a high fitting ability. However, this paper empirically observes that normalized Laplacian matrices frequently possess repeated eigenvalues. Moreover, we theoretically establish that the number of distinguishable eigenvalues plays a pivotal role in determining the expressive power of spectral graph neural networks. In light of this observation, we propose an eigenvalue correction strategy that can free polynomial filters from the constraints of repeated eigenvalue inputs. Concretely, the proposed eigenvalue correction strategy enhances the uniform distribution of eigenvalues, thus mitigating repeated eigenvalues, and improving the fitting capacity and expressive power of polynomial filters. Extensive experimental results on both synthetic and real-world datasets demonstrate the superiority of our method.

With the proliferation of dialogic data across the Internet, the Dialogue Commonsense Multi-choice Question Answering (DC-MCQ) task has emerged as a response to the challenge of comprehending user queries and intentions. Although prevailing methodologies exhibit effectiveness in addressing single-choice questions, they encounter difficulties in handling multi-choice queries due to the heightened intricacy and informational density. In this paper, inspired by the human cognitive process of progressively excluding options, we propose a three-step Reverse Exclusion Graph-of-Thought (ReX-GoT) framework, including Option Exclusion, Error Analysis, and Combine Information. Specifically, our ReX-GoT mimics human reasoning by gradually excluding irrelevant options and learning the reasons for option errors to choose the optimal path of the GoT and ultimately infer the correct answer. By progressively integrating intricate clues, our method effectively reduces the difficulty of multi-choice reasoning and provides a novel solution for DC-MCQ. Extensive experiments on the CICERO and CICERO_v2 datasets validate the significant improvement of our approach on DC-MCQ task. On zero-shot setting, our model outperform the best baseline by 17.67% in terms of F1 score for the multi-choice task. Most strikingly, our GPT3.5-based ReX-GoT framework achieves a remarkable 39.44% increase in F1 score.

Zhiyuan Liu, An Zhang, Hao Fei, Enzhi Zhang, Xiang Wang, Kenji Kawaguchi, Tat-Seng Chua. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

While the recent Chain-of-Thought (CoT) technique enhances the reasoning ability of large language models (LLMs) with the theory of mind, it might still struggle in handling logical reasoning that relies much on symbolic expressions and rigid deducing rules.To strengthen the logical reasoning capability of LLMs, we propose a novel Symbolic Chain-of-Thought, namely SymbCoT, a fully LLM-based framework that integrates symbolic expressions and logic rules with CoT prompting.Technically, building upon an LLM, SymbCoT 1) first translates the natural language context into the symbolic format, and then 2) derives a step-by-step plan to solve the problem with symbolic logical rules, 3) followed by a verifier to check the translation and reasoning chain.Via thorough evaluations on 5 standard datasets with both First-Order Logic and Constraint Optimization symbolic expressions, SymbCoT shows striking improvements over the CoT method consistently, meanwhile refreshing the current stateof-the-art performances.We further demonstrate that our system advances in more faithful, flexible, and explainable logical reasoning.To our knowledge, this is the first to combine symbolic expressions and rules into CoT for logical reasoning with LLMs.Code is open at https://github.com/Aiden0526/SymbCoT.

Structured Sentiment Analysis (SSA) was cast as a problem of bi-lexical dependency graph parsing by prior studies.Multiple formulations have been proposed to construct the graph, which share several intrinsic drawbacks: (1) The internal structures of spans are neglected, thus only the boundary tokens of spans are used for relation prediction and span recognition, thus hindering the model's expressiveness; (2) Long spans occupy a significant proportion in the SSA datasets, which further exacerbates the problem of internal structure neglect.In this paper, we treat the SSA task as a dependency parsing task on partially-observed dependency trees, regarding flat spans without determined tree annotations as latent subtrees to consider internal structures of spans.We propose a twostage parsing method and leverage TreeCRFs with a novel constrained inside algorithm to model latent structures explicitly, which also takes advantages of joint scoring graph arcs and headed spans for global optimization and inference.Results of extensive experiments on five benchmark datasets reveal that our method performs significantly better than all previous bi-lexical methods, achieving new state-of-theart.

Large Language Models (LLMs) are now being considered as judges of high efficiency to evaluate the quality of answers generated by candidate models. However, their judgments may be influenced by complex scenarios and inherent biases, raising concerns about their reliability. This study aims to bridge this gap by introducing four unexplored factors and examining the performance of LLMs as judges, namely answer quantity, inducing statements, judging strategy, and judging style. Additionally, we introduce a new dimension of question difficulty to provide a more comprehensive understanding of LLMs’ judgments across varying question intricacies. We employ ChatGPT, GPT-4, Gemini, and Claude-2 as judges and conduct experiments on Vicuna Benchmark and MT-bench. Our study reveals that LLMs’ judging abilities are susceptible to the influence of these four factors, and analyzing from the newly proposed dimension of question difficulty is highly necessary. We also provide valuable insights into optimizing LLMs’ performance as judges, enhancing their reliability and adaptability across diverse evaluation scenarios.

Text-to-video (T2V) synthesis has gained increasing attention in the community, in which the recently emerged diffusion models (DMs) have promisingly shown stronger performance than the past approaches. While existing state-of-the-art DMs are competent to achieve high-resolution video generation, they may largely suffer from key limitations (e.g., action occurrence disorders, crude video motions) with respect to the intricate temporal dynamics modeling, one of the crux of video synthesis. In this work, we investigate strengthening the awareness of video dynamics for DMs, for high-quality T2V generation. Inspired by human intuition, we design an innovative dynamic scene manager (dubbed as Dysen) module, which includes (step-1) extracting from input text the key actions with proper time-order arrangement, (step-2) transforming the action schedules into the dynamic scene graph (DSG) representations, and (step-3) enriching the scenes in the DSG with sufficient and reasonable details. Taking advantage of the existing powerful LLMs (e.g., ChatGPT) via in-context learning, Dysen realizes (nearly) human-level temporal dynamics understanding. Finally, the resulting video DSG with rich action scene details is encoded as fine-grained spatio-temporal features, integrated into the backbone T2V DM for video generating. Experiments on popular T2V datasets suggest that our Dysen- VDM consistently outperforms prior arts with significant margins, especially in scenarios with complex actions. Codes at http://haofei.vip/Dysen-VDM/.

Recent progress in Large Multimodal Models (LMM) has opened up great possibilities for various applications in the field of human-machine interactions. However, developing LMMs that can comprehend, reason, and plan in complex and diverse 3D environments remains a challenging topic, especially considering the demand for understanding permutation-invariant point cloud representations of the 3D scene. Existing works seek help from multi-view images by projecting 2D features to 3D space, which inevitably leads to huge computational overhead and performance degradation. In this paper, we present LL3DA, a Large Language 3D Assistant that takes point cloud as the direct input and responds to both text instructions and visual interactions. The additional visual interaction enables LMMs to better comprehend human interactions with the 3D environment and further remove the ambiguities within plain texts. Experiments show that LL3DA achieves remarkable results and surpasses various 3D vision-language models on both 3D Dense Captioning and 3D Question Answering.

In the rapidly evolving field of conversational AI, Ontology Expansion (OnExp) is crucial for enhancing the adaptability and robustness of conversational agents.Traditional models rely on static, predefined ontologies, limiting their ability to handle new and unforeseen user needs.This survey paper provides a comprehensive review of the state-of-the-art techniques in On-Exp for conversational understanding.It categorizes the existing literature into three main areas: (1) New Intent Discovery, (2) New Slot-Value Discovery, and (3) Joint OnExp.By examining the methodologies, benchmarks, and challenges associated with these areas, we highlight several emerging frontiers in OnExp to improve agent performance in real-world scenarios and discuss their corresponding challenges.This survey aspires to be a foundational reference for researchers and practitioners, promoting further exploration and innovation in this crucial domain.

In nested Named entity recognition (NER), entities are nested with each other, and thus requiring more data annotations to address. This leads to the development of few-shot nested NER, where the prevalence of pretrained language models with in-context learning (ICL) offers promising solutions. In this work, we introduce an effective and innovative ICL framework for the setting of few-shot nested NER. We improve the ICL prompt by devising a novel example demonstration selection mechanism, EnDe retriever. In EnDe retriever, we employ contrastive learning to perform three types of representation learning, in terms of semantic similarity, boundary similarity, and label similarity, to generate high-quality demonstration examples. Extensive experiments over three nested NER and four flat NER datasets demonstrate the efficacy of our system.

Existing research of video understanding still struggles to achieve in-depth comprehension and reasoning in complex videos, primarily due to the under-exploration of two key bottlenecks: fine-grained spatial-temporal perceptive understanding and cognitive-level video scene comprehension. This paper bridges the gap by presenting a novel solution. We first introduce a novel video Multimodal Large Language Model (MLLM), MotionEpic, which achieves fine-grained pixel-level spatial-temporal video grounding by integrating video spatial-temporal scene graph (STSG) representation. Building upon MotionEpic, we then develop a Video-of-Thought (VoT) reasoning framework. VoT inherits the Chain-of-Thought (CoT) core, breaking down a complex task into simpler and manageable sub-problems, and addressing them step-by-step from a low-level pixel perception to high-level cognitive interpretation. Extensive experiments across various complex video QA benchmarks demonstrate that our overall framework strikingly boosts existing state-of-the-art. To our knowledge, this is the first attempt at successfully implementing the CoT technique for achieving human-level video reasoning, where we show great potential in extending it to a wider range of video understanding scenarios. Systems and codes will be open later.

Large Language Models (LLMs) demonstrate remarkable proficiency in comprehending and handling text-based tasks. Many efforts are being made to transfer these attributes to video modality, which are termed Video-LLMs. However, existing Video-LLMs can only capture the coarse-grained semantics and are unable to effectively handle tasks related to comprehension or localization of specific video segments. In light of these challenges, we propose Momentor, a Video-LLM capable of accomplishing fine-grained temporal understanding tasks. To support the training of Momentor, we design an automatic data generation engine to construct Moment-10M, a large-scale video instruction dataset with segment-level instruction data. We train Momentor on Moment-10M, enabling it to perform segment-level reasoning and localization. Zero-shot evaluations on several tasks demonstrate that Momentor excels in fine-grained temporally grounded comprehension and localization.

While recently Multimodal Large Language Models (MM-LLMs) have made exciting strides, they mostly fall prey to the limitation of only input-side multimodal understanding, without the ability to produce content in multiple modalities. As we humans always perceive the world and communicate with people through various modalities, developing any-to-any MM-LLMs capable of accepting and delivering content in any modality becomes essential to human-level AI. To fill the gap, we present an end-to-end general-purpose any-to-any MM-LLM system, NExT-GPT. We connect an LLM with multimodal adaptors and different diffusion decoders, enabling NExT-GPT to perceive inputs and generate outputs in arbitrary combinations of text, image, video, and audio. By leveraging the existing well-trained high-performing encoders and decoders, NExT-GPT is tuned with only a small amount of parameter (1%) of certain projection layers, which not only benefits low-cost training but also facilitates convenient expansion to more potential modalities. Moreover, we introduce a modality-switching instruction tuning (MosIT) and manually curate a high-quality dataset for MosIT, based on which NExT-GPT is empowered with complex cross-modal semantic understanding and content generation. Overall, our research showcases the promising possibility of building a unified AI agent capable of modeling universal modalities, paving the way for more human-like AI research in the community.

Fact-checking fake news detection involves using verified accurate factual information in news reports as "evidence" to validate objective statement "claim". Existing works primarily focus on identifying critical elements within the evidence that support or refute specific claims by assessing the congruence or divergence between the claim and the associated evidence. These methods can broadly be divided into text-based and graph-based-the former centers on understanding the nuances of unstructured text to extract semantic word-level information. At the same time, the latter is proficient at analyzing the node-level structure of graphs it creates from the text to reveal topological insights. Each type provides a distinct view on identifying critical elements for fact-checking. To enhance the complementary nature of the two perspectives, this paper proposes an end-to-end framework for fact-checking fake news detection entitled Multi-view Counterfactual Contrastive Learning (MCCL). The framework incorporates a counterfactual technique to refine the fused features from both the "entity-view" of textual content and the "centrality-view" of the graph structure. Additionally, it employs contrastive learning to sharpen the distinctions among multi-view features, which facilitates the exact identification of critical elements in the evidence related to their respective claims. Experimental results on real datasets demonstrate that the proposed MCCL outperforms state-of-the-art methods.

Artificial intelligence (AI) encompasses knowledge acquisition and real-world grounding across various modalities, including language, visual, auditory, and sensory data. Multimodal large language models (MLLMs) have thus recently garnered growing interest in both academia and industry, showing an unprecedented trend to achieve human-level AI. This tutorial aims to deliver a comprehensive review of cutting-edge research in MLLMs, focusing on three key areas: MLLM architecture design, instructional learning, and multimodal reasoning of MLLMs. We will explore technical advancements, synthesize key challenges, and discuss potential avenues for future research. All the resources and materials will be made available online. https://mllm2024.github.io/ACM-MM2024

Visual programming, a modular paradigm, integrates different modules and Python operators to solve various vision-language tasks. Unlike end-to-end models that need task-specific data, it performs visual processing and inference in an unsupervised manner. Current visual programming methods generate programs in a single pass where the ability to evaluate and optimize based on feedback, unfortunately, is lacking, which consequentially limits their effectiveness for complex, multi-step problems. Inspired by benders decomposition, we introduce De-fine, a training-free framework that automatically decomposes complex tasks into simpler subtasks and refines programs through auto-feedback. This model-agnostic approach can improve logical reasoning performance by integrating the strengths of multiple models. Our experiments across various visual tasks show that De-fine creates more robust programs. Moreover, viewing each feedback module as an independent agent will yield fresh prospects for the field of agent research.

While existing Aspect-based Sentiment Analysis (ABSA) has received extensive effort and advancement, there are still gaps in defining a more holistic research target seamlessly integrating multimodality, conversation context, fine-granularity, and also covering the changing sentiment dynamics as well as cognitive causal rationales. This paper bridges the gaps by introducing a multimodal conversational ABSA, where two novel subtasks are proposed: 1) Panoptic Sentiment Sextuple Extraction, panoramically recognizing holder, target, aspect, opinion, sentiment, rationale from multi-turn multi-party multimodal dialogue. 2) Sentiment Flipping Analysis, detecting the dynamic sentiment transformation throughout the conversation with the causal reasons. To benchmark the tasks, we construct PanoSent, a dataset annotated both manually and automatically, featuring high quality, large scale, multimodality, multilingualism, multi-scenarios, and covering both implicit&explicit sentiment elements. To effectively address the tasks, we devise a novel Chain-of-Sentiment reasoning framework, together with a novel multimodal large language model (namely Sentica) and a paraphrase-based verification mechanism. Extensive evaluations demonstrate the superiority of our methods over strong baselines, validating the efficacy of all our proposed methods. The work is expected to open up a new era for the ABSA community, and thus all our codes and data are open at https://PanoSent.github.io/.

Event extraction (EE) is a critical direction in the field of information extraction, laying an important foundation for the construction of structured knowledge bases. EE from text has received ample research and attention for years, yet there can be numerous real-world applications that require direct information acquisition from speech signals, online meeting minutes, interview summaries, press releases, etc. While EE from speech has remained under-explored, this paper fills the gap by pioneering a SpeechEE, defined as detecting the event predicates and arguments from a given audio speech. To benchmark the SpeechEE task, we first construct a large-scale high-quality dataset. Based on textual EE datasets under the sentence, document, and dialogue scenarios, we convert texts into speeches through both manual real-person narration and automatic synthesis, empowering the data with diverse scenarios, languages, domains, ambiences, and speaker styles. Further, to effectively address the key challenges in the task, we tailor an E2E SpeechEE system based on the encoder-decoder architecture, where a novel Shrinking Unit module and a retrieval-aided decoding mechanism are devised. Extensive experimental results on all SpeechEE subsets demonstrate the efficacy of the proposed model, offering a strong baseline for the task. At last, being the first work on this topic, we shed light on key directions for future research. Our codes and the benchmark datasets are open at https://SpeechEE.github.io/

The Visual Spatial Description Challenge (VSD) is the first competition event focused on visual spatial understanding, organized under the auspices of the ACM Multimedia Conference 2024. The goal of the VSD challenge is to assess the the ability of models and systems to comprehend spatial concepts, relationships and other semantics from a scene presented with visual appearance. The VSD challenge provides two benchmark datasets for three subtasks, i.e., visual spatial relationship classification, single spatial description generation, and open-ended spatial description generation. The challenge details are available on https://lllogen.github.io/vsd-challenge.github.io/.

Coreference resolution, an essential task in natural language processing, is particularly challenging in multi-modal scenarios where data comes in various forms and modalities. Despite advancements, limitations due to scarce labeled data and underleveraged unlabeled data persist. We address these issues with a self-adaptive fine-grained multi-modal data augmentation framework for semi-supervised MCR, focusing on enriching training data from labeled datasets and tapping into the untapped potential of unlabeled data. Regarding the former issue, we first leverage text coreference resolution datasets and diffusion models,to perform fine-grained text-to-image generation with aligned text entities and image bounding boxes. We then introduce a self-adaptive selection strategy, meticulously curating the augmented data to enhance the diversity and volume of the training set without compromising its quality. For the latter issue, we design a self-adaptive threshold strategy that dynamically adjusts the confidence threshold based on the model's learning status and performance, enabling effective utilization of valuable information from unlabeled data. Additionally, we incorporate a distance smoothing term, which smooths distances between positive and negative samples, enhancing discriminative power of the model?s feature representations and addressing noise and uncertainty in the unlabeled data. Our experiments on the widely-used CIN dataset show that our framework significantly outperforms state-of-the-art baselines by at least 9.57% on MUC F1 score and 4.92% on CoNLL F1 score. Remarkably, against weakly-supervised baselines, our framework achieves a staggering 22.24% enhancement in MUC F1 score. These results, underpinned by in-depth analyses, underscore the effectiveness and potential of our approach for advancing MCR tasks.

Jinggui Liang, Lizi Liao, Hao Fei, Bobo Li, Jing Jiang. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Recent developments of vision large language models (LLMs) have seen remarkable progress, yet still encounter challenges towards multimodal generalists, such as coarse-grained instance-level understanding, lack of unified support for both images and videos, and insufficient coverage across various vision tasks. In this paper we present Vitron, a universal pixel-level vision LLM designed for comprehensive understanding, generating, segmenting, and editing of both static images and dynamic videos. Building on top of an LLM backbone, Vitron incorporates encoders for images, videos, and pixel-level regional visuals within its frontend modules, while employing state-of-the-art visual specialists as its backend, via which Vitron supports a spectrum of vision end tasks, spanning visual comprehension to visual generation, from low level to high level. To ensure an effective and precise message passing from LLM to backend modules for function invocation, we propose a novel hybrid method by simultaneously integrating discrete textual instructions and continuous signal embeddings. Further, we design various pixel-level spatiotemporal vision-language alignment learning for Vitron to reach the best fine-grained visual capability. Finally, a cross-task synergy module is advised to learn to maximize the task-invariant fine-grained visual features, enhancing the synergy between different visual tasks. Demonstrated over 12 visual tasks and evaluated across 22 datasets, Vitron showcases its extensive capabilities in the four main vision task clusters. Overall, this work illuminates the great potential of developing a more unified multimodal generalist.

In recent times, Vision-Language Models (VLMs) have been trained under two predominant paradigms. Generative training has enabled Multimodal Large Language Models (MLLMs) to tackle various complex tasks, yet issues such as hallucinations and weak object discrimination persist. Discriminative training, exemplified by models like CLIP, excels in zero-shot image-text classification and retrieval, yet struggles with complex scenarios requiring fine-grained semantic differentiation. This paper addresses these challenges by proposing a unified approach that integrates the strengths of both paradigms. Considering interleaved image-text sequences as the general format of input samples, we introduce a structure-induced training strategy that imposes semantic relationships between input samples and the MLLM’s hidden state. This approach enhances the MLLM’s ability to capture global semantics and distinguish fine-grained semantics. By leveraging dynamic sequence alignment within the Dynamic Time Warping framework and integrating a novel kernel for fine-grained semantic differentiation, our method effectively balances generative and discriminative tasks. Extensive experiments demonstrate the effectiveness of our approach, achieving state-of-the-art results in multiple generative tasks, especially those requiring cognitive and discrimination abilities. Additionally, our method surpasses discriminative benchmarks in interleaved and fine-grained retrieval tasks. By employing a retrieval-augmented generation strategy, our approach further enhances performance in some generative tasks within one model, offering a promising direction for future research in vision-language modeling.

The swift advancement in Multimodal LLMs (MLLMs) also presents significant challenges for effective knowledge editing. Current methods, including intrinsic knowledge editing and external knowledge resorting, each possess strengths and weaknesses, struggling to balance the desired properties of reliability, generality, and locality when applied to MLLMs. In this paper, we propose \textbf{UniKE}, a novel multimodal editing method that establishes a unified perspective and paradigm for intrinsic knowledge editing and external knowledge resorting. Both types of knowledge are conceptualized as vectorized key-value memories, with the corresponding editing processes resembling the assimilation and accommodation phases of human cognition, conducted at the same semantic levels. Within such a unified framework, we further promote knowledge collaboration by disentangling the knowledge representations into the semantic and truthfulness spaces. Extensive experiments validate the effectiveness of our method, which ensures that the post-edit MLLM simultaneously maintains excellent reliability, generality, and locality. The code for UniKE is available at https://github.com/beepkh/UniKE.

3D Referring Expression Segmentation (3D-RES) aims to segment 3D objects by correlating referring expressions with point clouds. However, traditional approaches frequently encounter issues like over-segmentation or mis-segmentation, due to insufficient emphasis on spatial information of instances. In this paper, we introduce a Rule-Guided Spatial Awareness Network (RG-SAN) by utilizing solely the spatial information of the target instance for supervision. This approach enables the network to accurately depict the spatial relationships among all entities described in the text, thus enhancing the reasoning capabilities. The RG-SAN consists of the Text-driven Localization Module (TLM) and the Rule-guided Weak Supervision (RWS) strategy. The TLM initially locates all mentioned instances and iteratively refines their positional information. The RWS strategy, acknowledging that only target objects have supervised positional information, employs dependency tree rules to precisely guide the core instance’s positioning. Extensive testing on the ScanRefer benchmark has shown that RG-SAN not only establishes new performance benchmarks, with an mIoU increase of 5.1 points, but also exhibits significant improvements in robustness when processing descriptions with spatial ambiguity. All codes are available at https://github.com/sosppxo/RG-SAN.

In this work, we propose a training-free method to inject visual prompts into Multimodal Large Language Models (MLLMs) through learnable latent variable optimization. We observe that attention, as the core module of MLLMs, connects text prompt tokens and visual tokens, ultimately determining the final results. Our approach involves adjusting visual tokens from the MLP output during inference, controlling the attention response to ensure text prompt tokens attend to visual tokens in referring regions. We optimize a learnable latent variable based on an energy function, enhancing the strength of referring regions in the attention map. This enables detailed region description and reasoning without the need for substantial training costs or model retraining. Our method offers a promising direction for integrating referring abilities into MLLMs, and supports referring with box, mask, scribble and point. The results demonstrate that our method exhibits out-of-domain generalization and interpretability.

Current universal segmentation methods demonstrate strong capabilities in pixel-level image and video understanding. However, they lack reasoning abilities and cannot be controlled via text instructions. In contrast, large vision-language multimodal models exhibit powerful vision-based conversation and reasoning capabilities but lack pixel-level understanding and have difficulty accepting visual prompts for flexible user interaction. This paper proposes OMG-LLaVA, a new and elegant framework combining powerful pixel-level vision understanding with reasoning abilities. It can accept various visual and text prompts for flexible user interaction. Specifically, we use a universal segmentation method as the visual encoder, integrating image information, perception priors, and visual prompts into visual tokens provided to the LLM. The LLM is responsible for understanding the user's text instructions and providing text responses and pixel-level segmentation results based on the visual information. We propose perception prior embedding to better integrate perception priors with image features. OMG-LLaVA achieves image-level, object-level, and pixel-level reasoning and understanding in a single model, matching or surpassing the performance of specialized methods on multiple benchmarks. Rather than using LLM to connect each specialist, our work aims at end-to-end training on one encoder, one decoder, and one LLM. The code and model have been released for further research.

In the visual spatial understanding (VSU) field, spatial image-to-text (SI2T) and spatial text-to-image (ST2I) are two fundamental tasks that appear in dual form. Existing methods for standalone SI2T or ST2I perform imperfectly in spatial understanding, due to the difficulty of 3D-wise spatial feature modeling. In this work, we consider modeling the SI2T and ST2I together under a dual learning framework. During the dual framework, we then propose to represent the 3D spatial scene features with a novel 3D scene graph (3DSG) representation that can be shared and beneficial to both tasks. Further, inspired by the intuition that the easier 3D$\to$image and 3D$\to$text processes also exist symmetrically in the ST2I and SI2T, respectively, we propose the Spatial Dual Discrete Diffusion (SD$^3$) framework, which utilizes the intermediate features of the 3D$\to$X processes to guide the hard X$\to$3D processes, such that the overall ST2I and SI2T will benefit each other. On the visual spatial understanding dataset VSD, our system outperforms the mainstream T2I and I2T methods significantly. Further in-depth analysis reveals how our dual learning strategy advances.

Recent studies indicate that dense retrieval models struggle to perform well on a wide variety of retrieval tasks that lack dedicated training data, as different retrieval tasks often entail distinct search intents. To address this challenge, in this work we leverage instructions to flexibly describe retrieval intents and introduce I3, a unified retrieval system that performs Intent-Introspective retrieval across various tasks, conditioned on Instructions without any task-specific training. I3 innovatively incorporates a pluggable introspector in a parameter-isolated manner to comprehend specific retrieval intents by jointly reasoning over the input query and instruction, and seamlessly integrates the introspected intent into the original retrieval model for intent-aware retrieval. Furthermore, we propose progressively-pruned intent learning. It utilizes extensive LLM-generated data to train I3 phase-by-phase, embodying two key designs: progressive structure pruning and drawback extrapolation-based data refinement. Extensive experiments show that in the BEIR benchmark, I3 significantly outperforms baseline methods designed with task-specific retrievers, achieving state-of-the-art zero-shot performance without any task-specific tuning.

With the increasing popularity of live streaming, the interactions from viewers during a live streaming can provide more specific and constructive feedback for both the streamer and platform. In such scenario, the primary and most direct feedback method from the audience is through comments. Thus, mining these live streaming comments to unearth the intentions behind them and, in turn, aiding streamers to enhance their live streaming quality is significant for the well development of live streaming ecosystem. To this end, we introduce the MMLSCU dataset, containing 50,129 intention-annotated comments across multiple modalities (text, images, vi-deos, audio) from eight streaming domains. Using multimodal pretrained large model and drawing inspiration from the Chain of Thoughts (CoT) concept, we implement an end-to-end model to sequentially perform the following tasks: viewer comment intent detection ➛ intent cause mining ➛ viewer comment explanation ➛ streamer policy suggestion. We employ distinct branches for video and audio to process their respective modalities. After obtaining the video and audio representations, we conduct a multimodal fusion with the comment. This integrated data is then fed into the large language model to perform inference across the four tasks following the CoT framework. Experimental results indicate that our model outperforms three multimodal classification baselines on comment intent detection and streamer policy suggestion, and one multimodal generation baselines on intent cause mining and viewer comment explanation. Compared to the models using only text, our multimodal setting yields superior outcomes. Moreover, incorporating CoT allows our model to enhance comment interpretation and more precise suggestions for the streamers. Our proposed dataset and model will bring new research attention on multimodal live streaming comment understanding.

Empathetic response generation is a crucial task in natural language processing , enabling emotionally resonant machine–human interactions. In this paper, we introduce the InfRa ( In tegrating Discourse F eatures and R esponse A ssessment) model to address limitations in traditional methods for this task, such as the lack of deep dialogue comprehension and response control. InfRa integrates discourse features to augment structural dialogue understanding, with a novel edge pruning and mutual information learning module to further refine the representation. The model also employs a response evaluation module for dynamic optimization , ensuring emotional and semantic consistency between the generated response and its context . Our experiments demonstrate that InfRa outperforms existing baselines, reducing the Perplexity (PPL) score by approximately 9 points and excelling in all three fine-grained aspects of human evaluation. This research not only advances the development of empathetic chatbots but also provides valuable insights for broader text generation tasks.

While pre-training large-scale video-language models (VLMs) has shown remarkable potential for various downstream video-language tasks, existing VLMs can still suffer from certain commonly seen limitations, e.g., coarse-grained cross-modal aligning, under-modeling of temporal dynamics, detached video-language view. In this work, we target enhancing VLMs with a fine-grained structural spatio-temporal alignment learning method (namely Finsta). First of all, we represent the input texts and videos with fine-grained scene graph (SG) structures, both of which are further unified into a holistic SG (HSG) for bridging two modalities. Then, an SG-based framework is built, where the textual SG (TSG) is encoded with a graph Transformer, while the video dynamic SG (DSG) and the HSG are modeled with a novel recurrent graph Transformer for spatial and temporal feature propagation. A spatial-temporal Gaussian differential graph Transformer is further devised to strengthen the sense of the changes in objects across spatial and temporal dimensions. Next, based on the fine-grained structural features of TSG and DSG, we perform object-centered spatial alignment and predicate-centered temporal alignment respectively, enhancing the video-language grounding in both the spatiality and temporality. We design our method as a plug&play system, which can be integrated into existing well-trained VLMs for further representation augmentation, without training from scratch or relying on SG annotations in downstream applications. On 6 representative VL modeling tasks over 12 datasets in both standard and long-form video scenarios, Finsta consistently improves the existing 13 strong-performing VLMs persistently, and refreshes the current state-of-the-art end task performance significantly in both the fine-tuning and zero-shot settings.

Few-shot named entity recognition (NER) exploits limited annotated instances to identify named mentions. Effectively transferring the internal or external resources thus becomes the key to few-shot NER. While the existing prompt tuning methods have shown remarkable few-shot performances, they still fail to make full use of knowledge. In this work, we investigate the integration of rich knowledge to prompt tuning for stronger few-shot NER. We propose incorporating the deep prompt tuning framework with threefold knowledge (namelyTKDP), including the internal 1)context knowledgeand the external 2)label knowledge& 3)sememe knowledge. TKDP encodes the three feature sources and incorporates them into soft prompt embeddings, which are further injected into an existing pre-trained language model to facilitate predictions. On five benchmark datasets, the performance of our knowledge-enriched model was boosted by at most 11.53% F1 over the raw deep prompt method, and it significantly outperforms 9 strong-performing baseline systems in 5-/10-/20-shot settings, showing great potential in few-shot NER. Our TKDP framework can be broadly adapted to other few-shot tasks without much effort.

In this work, we investigate a more realistic unsupervised multimodal machine translation (UMMT) setup, inference-time image-free UMMT, where the model is trained with sourcetext image pairs, and tested with only sourcetext inputs.First, we represent the input images and texts with the visual and language scene graphs (SG), where such fine-grained visionlanguage features ensure a holistic understanding of the semantics.To enable pure-text input during inference, we devise a visual scene hallucination mechanism that dynamically generates pseudo visual SG from the given textual SG.Several SG-pivoting based learning objectives are introduced for unsupervised translation training.On the benchmark Multi30K data, our SG-based method outperforms the best-performing baseline by significant BLEU scores on the task and setup, helping yield translations with better completeness, relevance and fluency without relying on paired images.Further in-depth analyses reveal how our model advances in the task setting.

Existing research on multimodal relation extraction (MRE) faces two co-existing challenges, internal-information over-utilization and external-information under-exploitation.To combat that, we propose a novel framework that simultaneously implements the idea of internal-information screening and externalinformation exploiting.First, we represent the fine-grained semantic structures of the input image and text with the visual and textual scene graphs, which are further fused into a unified cross-modal graph (CMG).Based on CMG, we perform structure refinement with the guidance of the graph information bottleneck principle, actively denoising the less-informative features.Next, we perform topic modeling over the input image and text, incorporating latent multimodal topic features to enrich the contexts.On the benchmark MRE dataset, our system outperforms the current best model significantly.With further in-depth analyses, we reveal the great potential of our method for the MRE task.Our codes are open at https://github.com/ChocoWu/MRE-ISE.

Unpaired cross-lingual image captioning has long suffered from irrelevancy and disfluency issues, due to the inconsistencies of the semantic scene and syntax attributes during transfer.In this work, we propose to address the above problems by incorporating the scene graph (SG) structures and the syntactic constituency (SC) trees.Our captioner contains the semantic structure-guided image-to-pivot captioning and the syntactic structure-guided pivot-to-target translation, two of which are joined via pivot language.We then take the SG and SC structures as pivoting, performing cross-modal semantic structure alignment and cross-lingual syntactic structure alignment learning.We further introduce cross-lingual&cross-modal back-translation training to fully align the captioning and translation stages.Experiments on English↔Chinese transfers show that our model shows great superiority in improving captioning relevancy and fluency.

Visual spatial description (VSD) aims to generate texts that describe the spatial relations of the given objects within images.Existing VSD work merely models the 2D geometrical vision features, thus inevitably falling prey to the problem of skewed spatial understanding of target objects.In this work, we investigate the incorporation of 3D scene features for VSD.With an external 3D scene extractor, we obtain the 3D objects and scene features for input images, based on which we construct a target object-centered 3D spatial scene graph (GO3D-S 2 G), such that we model the spatial semantics of target objects within the holistic 3D scenes.Besides, we propose a scene subgraph selecting mechanism, sampling topologically-diverse subgraphs from GO3D-S 2 G, where the diverse local structure features are navigated to yield spatially-diversified text generation.Experimental results on two VSD datasets demonstrate that our framework outperforms the baselines significantly, especially improving on the cases with complex visual spatial relations.Meanwhile, our method can produce more spatially-diversified generation.Code is available at https://github.com/zhaoyucs/VSD.

Language Models (LMs) have demonstrated impressive molecule understanding ability on various 1D text-related tasks.However, they inherently lack 2D graph perception -a critical ability of human professionals in comprehending molecules' topological structures.To bridge this gap, we propose MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter.MolCA enables an LM (i.e., Galactica) to understand both text-and graph-based molecular contents via the cross-modal projector.Specifically, the cross-modal projector is implemented as a Q-Former to connect a graph encoder's representation space and an LM's text space.Further, MolCA employs a uni-modal adapter (i.e., LoRA) for the LM's efficient adaptation to downstream tasks.Unlike previous studies that couple an LM with a graph encoder via cross-modal contrastive learning, MolCA retains the LM's ability of open-ended text generation and augments it with 2D graph information.To showcase its effectiveness, we extensively benchmark MolCA on tasks of molecule captioning, IUPAC name prediction, and molecule-text retrieval, on which MolCA significantly outperforms the baselines.Our codes and checkpoints can be found at https: //github.com/acharkq/MolCA.

Given a descriptive language query, Video Moment Retrieval (VMR) aims to seek the corresponding semantic-consistent moment clip in the video, which is represented as a pair of the start and end timestamps. Although current methods have achieved satisfying performance, training these models heavily relies on the fully-annotated VMR datasets. Nonetheless, precise video temporal annotations are extremely labor-intensive and ambiguous due to the diverse preferences of different annotators.

Information retrieval (IR) is a fundamental technique that aims to acquire information from a collection of documents, web pages, or other sources. While traditional text-based IR has achieved great success, the under-utilization of varied data sources in different modalities (i.e., text, images, audio, and video) would hinder IR techniques from giving its full advancement and thus limits the applications in the real world. Within recent years, the rapid development of deep multimodal learning paves the way for advancing IR with multi-modality. Benefiting from a variety of data types and modalities, some latest prevailing techniques are invented to show great facilitation in multi-modal and IR learning, such as CLIP, ChatGPT, GPT4, etc. In the context of IR, deep multi-modal learning has shown the prominent potential to improve the performance of retrieval systems, by enabling them to better understand and process the diverse types of data that they encounter. Given the great potential shown by multimodal-empowered IR, there can be still unsolved challenges and open questions in the related directions. With this workshop, we aim to provide a platform for discussion about multi-modal IR among scholars, practitioners, and other interested parties.

It has been a hot research topic to enable machines to understand human emotions in multimodal contexts under dialogue scenarios, which is tasked with multimodal emotion analysis in conversation (MM-ERC). MM-ERC has received consistent attention in recent years, where a diverse range of methods has been proposed for securing better task performance. Most existing works treat MM-ERC as a standard multimodal classification problem and perform multimodal feature disentanglement and fusion for maximizing feature utility. Yet after revisiting the characteristic of MM-ERC, we argue that both the feature multimodality and conversational contextualization should be properly modeled simultaneously during the feature disentanglement and fusion steps. In this work, we target further pushing the task performance by taking full consideration of the above insights. On the one hand, during feature disentanglement, based on the contrastive learning technique, we devise a Dual-level Disentanglement Mechanism (DDM) to decouple the features into both the modality space and utterance space. On the other hand, during the feature fusion stage, we propose a Contribution-aware Fusion Mechanism (CFM) and a Context Refusion Mechanism (CRM) for multimodal and context integration, respectively. They together schedule the proper integrations of multimodal and context features. Specifically, CFM explicitly manages the multimodal feature contributions dynamically, while CRM flexibly coordinates the introduction of dialogue contexts. On two public MM-ERC datasets, our system achieves new state-of-the-art performance consistently. Further analyses demonstrate that all our proposed mechanisms greatly facilitate the MM-ERC task by making full use of the multimodal and context features adaptively. Note that our proposed methods have the great potential to facilitate a broader range of other conversational multimodal tasks.

In the text-to-image generation field, recent remarkable progress in Stable Diffusion makes it possible to generate rich kinds of novel photorealistic images. However, current models still face misalignment issues (e.g., problematic spatial relation understanding and numeration failure) in complex natural scenes, which impedes the high-faithfulness text-to-image generation. Although recent efforts have been made to improve controllability by giving fine-grained guidance (e.g., sketch and scribbles), this issue has not been fundamentally tackled since users have to provide such guidance information manually. In this work, we strive to synthesize high-fidelity images that are semantically aligned with a given textual prompt without any guidance. Toward this end, we propose a coarse-to-fine paradigm to achieve layout planning and image generation. Concretely, we first generate the coarse-grained layout conditioned on a given textual prompt via in-context learning based on Large Language Models. Afterward, we propose a fine-grained object-interaction diffusion method to synthesize high-faithfulness images conditioned on the prompt and the automatically generated layout. Extensive experiments demonstrate that our proposed method outperforms the state-of-the-art models in terms of layout and image generation. Our code and settings are available at https://layoutllm-t2i.github.io/.

As one of the core video semantic understanding tasks, Video Semantic Role Labeling (VidSRL) aims to detect the salient events from given videos, by recognizing the predict-argument event structures and the interrelationships between events. While recent endeavors have put forth methods for VidSRL, they can be mostly subject to two key drawbacks, including the lack of fine-grained spatial scene perception and the insufficiently modeling of video temporality. Towards this end, this work explores a novel holistic spatio-temporal scene graph (namely HostSG) representation based on the existing dynamic scene graph structures, which well model both the fine-grained spatial semantics and temporal dynamics of videos for VidSRL. Built upon the HostSG, we present a nichetargeting VidSRL framework. A scene-event mapping mechanism is first designed to bridge the gap between the underlying scene structure and the high-level event semantic structure, resulting in an overall hierarchical scene-event (termed ICE) graph structure. We further perform iterative structure refinement to optimize the ICE graph, e.g., filtering noisy branches and newly building informative connections, such that the overall structure representation can best coincide with end task demand. Finally, three subtask predictions of VidSRL are jointly decoded, where the end-to-end paradigm effectively avoids error propagation. On the benchmark dataset, our framework boosts significantly over the current best-performing model. Further analyses are shown for a better understanding of the advances of our methods. Our HostSG representation shows greater potential to facilitate a broader range of other video understanding tasks.

In this work, we investigate the task of text-to-image (T2I) synthesis under the abstract-to-intricate setting, i.e., generating intricate visual content from simple abstract text prompts. Inspired by human imagination intuition, we propose a novel scene-graph hallucination (SGH) mechanism for effective abstract-to-intricate T2I synthesis. SGH carries out scene hallucination by expanding the initial scene graph (SG) of the input prompt with more feasible specific scene structures, in which the structured semantic representation of SG ensures high controllability of the intrinsic scene imagination. To approach the T2I synthesis, we deliberately build an SG-based hallucination diffusion system. First, we implement the SGH module based on the discrete diffusion technique, which evolves the SG structure by iteratively adding new scene elements. Then, we utilize another continuous-state diffusion model as the T2I synthesizer, where the overt image-generating process is navigated by the underlying semantic scene structure induced from the SGH module. On the benchmark COCO dataset, our system outperforms the existing best-performing T2I model by a significant margin, especially improving on the abstract-to-intricate T2I generation. Further in-depth analyses reveal how our methods advance.

Since developing a new multimodal LLM (MLLM) by pre-training on tremendous image-text pairs from scratch can be exceedingly resource-consuming, connecting an existing LLM with a comparatively lightweight visual prompt generator (VPG) becomes a feasible paradigm. However, further tuning the VPG component of the MLLM still incurs significant computational costs, such as thousands of GPU hours and millions of training data points. An alternative solution is transferring an existing VPG from one MLLM to the target MLLM. In this work, we investigate VPG transferability across LLMs for the first time, aiming to reduce the cost of VPG training. Specifically, we explore VPG transfer across different LLM sizes (e.g., small-to-large) and types. We identify key factors to maximize transfer efficiency, based on which we develop a simple yet highly effective two-stage transfer framework, called VPGTrans. Notably, it enables VPG transfer from BLIP-2 OPT 2.7B to BLIP-2 OPT 6.7B with less than 10% of the GPU hours using only 10.7% of the training data compared to training a VPG for OPT 6.7B from scratch. Furthermore, we provide a series of intriguing findings and discuss potential explanations behind them. Finally, we showcase the practical value of our VPGTrans approach, by customizing two novel MLLMs, including VL-LLaMA and VL-Vicuna, with recently released LLaMA and Vicuna LLMs.

Aspect-based sentiment triplet extraction (ASTE) aims at recognizing the joint triplets from texts, i.e., aspect terms, opinion expressions, and correlated sentiment polarities. As a newly proposed task, ASTE depicts the complete sentiment picture from different perspectives to better facilitate real-world applications. Unfortunately, several major challenges, such as the overlapping issue and long-distance dependency, have not been addressed effectively by the existing ASTE methods, which limits the performance of the task. In this article, we present an innovative encoder-decoder framework for end-to-end ASTE. Specifically, the ASTE task is first modeled as an unordered triplet set prediction problem, which is satisfied with a nonautoregressive decoding paradigm with a pointer network. Second, a novel high-order aggregation mechanism is proposed for fully integrating the underlying interactions between the overlapping structure of aspect and opinion terms. Third, a bipartite matching loss is introduced for facilitating the training of our nonautoregressive system. Experimental results on benchmark datasets show that our proposed framework significantly outperforms the state-of-the-art methods. Further analysis demonstrates the advantages of the proposed framework in handling the overlapping issue, relieving long-distance dependency and decoding efficiency.

Aspect-based sentiment analysis (ABSA) aims at automatically inferring the specific sentiment polarities toward certain aspects of products or services behind the social media texts or reviews, which has been a fundamental application to the real-world society. Since the early 2010s, ABSA has achieved extraordinarily high accuracy with various deep neural models. However, existing ABSA models with strong in-house performances may fail to generalize to some challenging cases where the contexts are variable, i.e., low robustness to real-world environments. In this study, we propose to enhance the ABSA robustness by systematically rethinking the bottlenecks from all possible angles, including model, data, and training. First, we strengthen the current best-robust syntax-aware models by further incorporating the rich external syntactic dependencies and the labels with aspect simultaneously with a universal-syntax graph convolutional network. In the corpus perspective, we propose to automatically induce high-quality synthetic training data with various types, allowing models to learn sufficient inductive bias for better robustness. Last, we based on the rich pseudo data perform adversarial training to enhance the resistance to the context perturbation and meanwhile employ contrastive learning to reinforce the representations of instances with contrastive sentiments. Extensive robustness evaluations are conducted. The results demonstrate that our enhanced syntax-aware model achieves better robustness performances than all the state-of-the-art baselines. By additionally incorporating our synthetic corpus, the robust testing results are pushed with around 10% accuracy, which are then further improved by installing the advanced training strategies. In-depth analyses are presented for revealing the factors influencing the ABSA robustness.

So far, named entity recognition (NER) has been involved with three major types, including flat, overlapped (aka. nested), and discontinuous NER, which have mostly been studied individually. Recently, a growing interest has been built for unified NER, tackling the above three jobs concurrently with one single model. Current best-performing methods mainly include span-based and sequence-to-sequence models, where unfortunately the former merely focus on boundary identification and the latter may suffer from exposure bias. In this work, we present a novel alternative by modeling the unified NER as word-word relation classification, namely W^2NER. The architecture resolves the kernel bottleneck of unified NER by effectively modeling the neighboring relations between entity words with Next-Neighboring-Word (NNW) and Tail-Head-Word-* (THW-*) relations. Based on the W^2NER scheme we develop a neural framework, in which the unified NER is modeled as a 2D grid of word pairs. We then propose multi-granularity 2D convolutions for better refining the grid representations. Finally, a co-predictor is used to sufficiently reason the word-word relations. We perform extensive experiments on 14 widely-used benchmark datasets for flat, overlapped, and discontinuous NER (8 English and 6 Chinese datasets), where our model beats all the current top-performing baselines, pushing the state-of-the-art performances of unified NER.

Unified opinion role labeling (ORL) aims to detect all possible opinion structures of 'opinion-holder-target' in one shot, given a text. The existing transition-based unified method, unfortunately, is subject to longer opinion terms and fails to solve the term overlap issue. Current top performance has been achieved by employing the span-based graph model, which however still suffers from both high model complexity and insufficient interaction among opinions and roles. In this work, we investigate a novel solution by revisiting the transition architecture, and augmenting it with a pointer network (PointNet). The framework parses out all opinion structures in linear-time complexity, meanwhile breaks through the limitation of any length of terms with PointNet. To achieve the explicit opinion-role interactions, we further propose a unified dependency-opinion graph (UDOG), co-modeling the syntactic dependency structure and the partial opinion-role structure. We then devise a relation-centered graph aggregator (RCGA) to encode the multi-relational UDOG, where the resulting high-order representations are used to promote the predictions in the vanilla transition system. Our model achieves new state-of-the-art results on the MPQA benchmark. Analyses further demonstrate the superiority of our methods on both efficacy and efficiency.

The state-of-the-art model for structured sentiment analysis casts the task as a dependency parsing problem, which has some limitations:(1) The label proportions for span prediction and span relation prediction are imbalanced.(2) The span lengths of sentiment tuple components may be very large in this task, which will further exacerbates the imbalance problem.(3) Two nodes in a dependency graph cannot have multiple arcs, therefore some overlapped sentiment tuples cannot be recognized.In this work, we propose nichetargeting solutions for these issues.First, we introduce a novel labeling strategy, which contains two sets of token pair labels, namely essential label set and whole label set.The essential label set consists of the basic labels for this task, which are relatively balanced and applied in the prediction layer.The whole label set includes rich labels to help our model capture various token relations, which are applied in the hidden layer to softly influence our model.Moreover, we also propose an effective model to well collaborate with our labeling strategy, which is equipped with the graph attention networks to iteratively refine token representations, and the adaptive multi-label classifier to dynamically predict multiple relations between token pairs.We perform extensive experiments on 5 benchmark datasets in four languages.Experimental results show that our model outperforms previous SOTA models by a large margin. 1

Event extraction (EE) is an essential task of information extraction, which aims to extract structured event information from unstructured text. Most prior work focuses on extracting flat events while neglecting overlapped or nested ones. A few models for overlapped and nested EE includes several successive stages to extract event triggers and arguments,which suffer from error propagation. Therefore, we design a simple yet effective tagging scheme and model to formulate EE as word-word relation recognition, called OneEE. The relations between trigger or argument words are simultaneously recognized in one stage with parallel grid tagging, thus yielding a very fast event extraction speed. The model is equipped with an adaptive event fusion module to generate event-aware representations and a distance-aware predictor to integrate relative distance information for word-word relation recognition, which are empirically demonstrated to be effective mechanisms. Experiments on 3 overlapped and nested EE benchmarks, namely FewFC, Genia11, and Genia13, show that OneEE achieves the state-of-the-art (SOTA) results. Moreover, the inference speed of OneEE is faster than those of baselines in the same condition, and can be further substantially improved since it supports parallel inference.

Emotion cause pair extraction (ECPE), as one of the derived subtasks of emotion cause analysis (ECA), shares rich inter-related features with emotion extraction (EE) and cause extraction (CE). Therefore EE and CE are frequently utilized as auxiliary tasks for better feature learning, modeled via multi-task learning (MTL) framework by prior works to achieve state-of-the-art (SoTA) ECPE results. However, existing MTL-based methods either fail to simultaneously model the specific features and the interactive feature in between, or suffer from the inconsistency of label prediction. In this work, we consider addressing the above challenges for improving ECPE by performing two alignment mechanisms with a novel Aˆ2Net model. We first propose a feature-task alignment to explicitly model the specific emotion-&cause-specific features and the shared interactive feature. Besides, an inter-task alignment is implemented, in which the label distance between the ECPE and the combinations of EE&CE are learned to be narrowed for better label consistency. Evaluations of benchmarks show that our methods outperform current best-performing systems on all ECA subtasks. Further analysis proves the importance of our proposed alignment mechanisms for the task.

Relation Extraction (RE) is a fundamental task of information extraction, which has attracted a large amount of research attention.Previous studies focus on extracting the relations within a sentence or document, while currently researchers begin to explore cross-document RE.However, current cross-document RE methods directly utilize text snippets surrounding the target entities in multiple given documents, which brings considerable noisy and non-relevant sentences.Moreover, they utilize all the text paths in a document bag in a coarse-grained way, without considering the connections between these text paths.In this paper, we aim to address both of these shortages and push the stateof-the-art for cross-document RE.First, we focus on input construction for our RE model and propose an entity-based document-context filter to retain useful information in the given documents by using the bridge entities in the text paths.Second, we propose a cross-document RE model based on cross-path entity relation attention, which allows the entity relations across text paths to interact with each other.We compare our cross-document RE method with the state-of-the-art methods in the dataset CodRED.Our method outperforms them by at least 10% in F1, thus demonstrating its effectiveness.

Many natural language processing (NLP) tasks appear in dual forms, which are generally solved by dual learning technique that models the dualities between the coupled tasks. In this work, we propose to further enhance dual learning with structure matching that explicitly builds structural connections in between. Starting with the dual text$\leftrightarrow$text generation, we perform dually-syntactic structure co-echoing of the region of interest (RoI) between the task pair, together with a syntax cross-reconstruction at the decoding side. We next extend the idea to a text$\leftrightarrow$non-text setup, making alignment between the syntactic-semantic structure. Over 2*14 tasks covering 5 dual learning scenarios, the proposed structure matching method shows its significant effectiveness in enhancing existing dual learning. Our method can retrieve the key RoIs that are highly crucial to the task performance. Besides NLP tasks, it is also revealed that our approach has great potential in facilitating more non-text$\leftrightarrow$non-text scenarios.

So far, aspect-based sentiment analysis (ABSA) has involved with total seven subtasks, in which, however the interactions among them have been left unexplored sufficiently. This work presents a novel multiplex cascade framework for unified ABSA and maintaining such interactions. First, we model total seven subtasks as a hierarchical dependency in the easy-to-hard order, based on which we then propose a multiplex decoding mechanism, transferring the sentiment layouts and clues in lower tasks to upper ones. The multiplex strategy enables highly-efficient subtask interflows and avoids repetitive training; meanwhile it sufficiently utilizes the existing data without requiring any further annotation. Further, based on the characteristics of aspect-opinion term extraction and pairing, we enhance our multiplex framework by integrating POS tag and syntactic dependency information for term boundary and pairing identification. The proposed Syntax-aware Multiplex (SyMux) framework enhances the ABSA performances on 28 subtasks (7×4 datasets) with big margins.

Recent research attention for relation extraction has been paid to the dialogue scenario, i.e., dialogue-level relation extraction (DiaRE). Existing DiaRE methods either simply concatenate the utterances in a dialogue into a long piece of text, or employ naive words, sentences or entities to build dialogue graphs, while the structural characteristics in dialogues have not been fully utilized. In this work, we investigate a novel dialogue-level mixed dependency graph (D2G) and an argument reasoning graph (ARG) for DiaRE with a global relation reasoning mechanism. First, we model the entire dialogue into a unified and coherent D2G by explicitly integrating both syntactic and discourse structures, which enables richer semantic and feature learning for relation extraction. Second, we stack an ARG graph on top of D2G to further focus on argument inter-dependency learning and argument representation refinement, for sufficient argument relation inference. In our global reasoning framework, D2G and ARG work collaboratively, iteratively performing lexical, syntactic and semantic information exchange and representation learning over the entire dialogue context. On two DiaRE benchmarks, our framework shows considerable improvements over the current state-of-the-art baselines. Further analyses show that the model effectively solves the long-range dependence issue, and meanwhile gives explainable predictions.

Conversational semantic role labeling (CSRL) is a newly proposed task that uncovers the shallow semantic structures in a dialogue text. Unfortunately several important characteristics of the CSRL task have been overlooked by the existing works, such as the structural information integration, near-neighbor influence. In this work, we investigate the integration of a latent graph for CSRL. We propose to automatically induce a predicate-oriented latent graph (POLar) with a predicate-centered gaussian mechanism, by which the nearer and informative words to the predicate will be allocated with more attention. The POLar structure is then dynamically pruned and refined so as to best fit the task need. We additionally introduce an effective dialogue-level pre-trained language model, CoDiaBERT, for better supporting multiple utterance sentences and handling the speaker coreference issue in CSRL. Our system outperforms best-performing baselines on three benchmark CSRL datasets with big margins, especially achieving over 4% F1 score improvements on the cross-utterance argument detection. Further analyses are presented to better understand the effectiveness of our proposed methods.

Universally modeling all typical information extraction tasks (UIE) with one generative language model (GLM) has revealed great potential by the latest study, where various IE predictions are unified into a linearized hierarchical expression under a GLM. Syntactic structure information, a type of effective feature which has been extensively utilized in IE community, should also be beneficial to UIE. In this work, we propose a novel structure-aware GLM, fully unleashing the power of syntactic knowledge for UIE. A heterogeneous structure inductor is explored to unsupervisedly induce rich heterogeneous structural representations by post-training an existing GLM. In particular, a structural broadcaster is devised to compact various latent trees into explicit high-order forests, helping to guide a better generation during decoding. We finally introduce a task-oriented structure fine-tuning mechanism, further adjusting the learned structures to most coincide with the end-task's need. Over 12 IE benchmarks across 7 tasks our system shows significant improvements over the baseline UIE system. Further in-depth analyses show that our GLM learns rich task-adaptive structural bias that greatly resolves the UIE crux, the long-range dependence issue and boundary identifying.

Fine-grained sentiment classification (FGSC) task and fine-grained controllable text generation (FGSG) task are two representative applications of sentiment analysis, two of which together can actually form an inverse task prediction, i.e., the former aims to infer the fine-grained sentiment polarities given a text piece, while the latter generates text content that describes the input fine-grained opinions. Most of the existing work solves the FGSC and the FGSG tasks in isolation, while ignoring the complementary benefits in between. This paper combines FGSC and FGSG as a joint dual learning system, encouraging them to learn the advantages from each other. Based on the dual learning framework, we further propose decoupling the feature representations in two tasks into fine-grained aspect-oriented opinion variables and content variables respectively, by performing mutual disentanglement learning upon them. We also propose to transform the difficult "data-to-text'' generation fashion widely used in FGSG into an easier text-to-text generation fashion by creating surrogate natural language text as the model inputs. Experimental results on 7 sentiment analysis benchmarks including both the document-level and sentence-level datasets show that our method significantly outperforms the current strong-performing baselines on both the FGSC and FGSG tasks. Automatic and human evaluations demonstrate that our FGSG model successfully generates fluent, diverse and rich content conditioned on fine-grained sentiments.

Joint aspect category sentiment analysis (ACSA) and rating prediction (RP) is a newly proposed task (namely ASAP) that integrates the characteristics of both fine-grained and coarse-grained sentiment analysis. However, the prior joint models for the ASAP task only consider the shallow interaction between the two granularities. In this work, we gain the inspiration from human intuition, presenting an innovative from-fine-to-coarse reasoning framework for better joint task performance. Our system advances mainly in three aspects. First, we additionally make use of the category label text features, co-encoding them with the input document texts, allowing to accurately capture the key clues of each category. Second, we build a fine-to-coarse hierarchical label graph, modeling the aspect categories and the overall rating as a hierarchical structure for full interaction of the two granularities. Third, we propose to perform global iterative reasoning with a cross-collaboration between the hierarchical label graph and the context graphs, enabling sufficient communication between categories and review contexts. Based on the ASAP dataset, experimental results demonstrate that our proposed framework outperforms state-of-the-art baselines by large margins. Further in-depth analyses prove that our method is effective on addressing both the unbalanced data distribution and the long-text issue.

Attention has been shown highly effective for modeling sequences, capturing the more informative parts in learning a deep representation. However, recent studies show that the attention values do not always coincide with intuition in tasks, such as machine translation and sentiment classification. In this study, we consider using deep reinforcement learning to automatically optimize attention distribution during the minimization of end task training losses. With more sufficient environment states, iterative actions are taken to adjust attention weights so that more informative words receive more attention automatically. Results on different tasks and different attention networks demonstrate that our model is of great effectiveness in improving the end task performances, yielding more reasonable attention distribution. The more in-depth analysis further reveals that our retrofitting method can help to bring explainability for baseline attention.

A majority of research interests in irregular (e.g., nested or discontinuous) named entity recognition (NER) have been paid on nested entities, while discontinuous entities received limited attention. Existing work for discontinuous NER, however, either suffers from decoding ambiguity or predicting using token-level local features. In this work, we present an innovative model for discontinuous NER based on pointer networks, where the pointer simultaneously decides whether a token at each decoding frame constitutes an entity mention and where the next constituent token is. Our model has three major merits compared with previous work: (1) The pointer mechanism is memory-augmented, which enhances the mention boundary detection and interactions between the current decision and prior recognized mentions. (2) The encoder-decoder architecture can linearize the complexity of structure prediction, and thus reduce search costs. (3) The model makes every decision using global information, i.e., by consulting all the input, encoder and previous decoder output in a global view. Experimental results on the CADEC and ShARe13 datasets show that our model outperforms flat and hypergraph models as well as a state-of-the-art transition-based model for discontinuous NER. Further in-depth analysis demonstrates that our model performs well in recognizing various entities including flat, overlapping and discontinuous ones. More crucially, our model is effective on boundary detection, which is the kernel source to NER.

Currently the unified semantic role labeling (SRL) that achieves predicate identification and argument role labeling in an end-to-end manner has received growing interests. Recent works show that leveraging the syntax knowledge significantly enhances the SRL performances. In this paper, we investigate a novel unified SRL framework based on the sequence-to-sequence architecture with double enhancement in both the encoder and decoder sides. In the encoder side, we propose a novel label-aware graph convolutional network (LA-GCN) to encode both the syntactic dependent arcs and labels into BERT-based word representations. In the decoder side, we creatively design a pointer-network-based model for detecting predicates, arguments and roles jointly. Our pointer-net decoder is able to make decisions by consulting all the input elements in a global view, and meanwhile it is syntactic-aware by incorporating the syntax information from LA-GCN. Besides, a high-order interacted attention is introduced into the decoder for leveraging previously recognized triplets to help the current decision. Empirical experiments show that our framework significantly outperforms all existing graph-based methods on the CoNLL09 and Universal Proposition Bank datasets. In-depth analysis demonstrates that our model can effectively capture the correlations between syntactic and SRL structures.

End-to-end semantic role labeling (SRL) has been received increasing interest. It performs the two subtasks of SRL: predicate identification and argument role labeling, jointly. Recent work is mostly focused on graph-based neural models, while the transition-based framework with neural networks which has been widely used in a number of closely-related tasks, has not been studied for the joint task yet. In this paper, we present the first work of transition-based neural models for end-to-end SRL. Our transition model incrementally discovers all sentential predicates as well as their arguments by a set of transition actions. The actions of the two subtasks are executed mutually for full interactions. Besides, we suggest high-order compositions to extract non-local features, which can enhance the proposed transition model further. Experimental results on CoNLL09 and Universal Proposition Bank show that our final model can produce state-of-the-art performance, and meanwhile keeps highly efficient in decoding. We also conduct detailed experimental analysis for a deep understanding of our proposed model.

In this paper, we propose to enhance the pair-wise aspect and opinion terms extraction (PAOTE) task by incorporating rich syntactic knowledge. We first build a syntax fusion encoder for encoding syntactic features, including a label-aware graph convolutional network (LAGCN) for modeling the dependency edges and labels, as well as the POS tags unifiedly, and a local-attention module encoding POS tags for better term boundary detection. During pairing, we then adopt Biaffine and Triaffine scoring for high-order aspect-opinion term pairing, in the meantime re-harnessing the syntax-enriched representations in LAGCN for syntactic-aware scoring. Experimental results on four benchmark datasets demonstrate that our model outperforms current state-of-the-art baselines, meanwhile yielding explainable predictions with syntactic knowledge.

Existing works for document-level sentiment classification task treat the review document as an overall text unit, performing feature extraction with various sophisticated model architectures. In this paper, we draw inspiration from fine-grained sentiment analysis, proposing to first learn the latent target-opinion distribution behind the documents, and then leverage such fine-grained prior knowledge into the classification process. We model the latent target-opinion distribution as hierarchical variables, where global-level variable captures the overall target and opinion, and local-level variables retrieve the detailed opinion clues at the word level. The proposed method consists of two main parts: a variational module and a classification module. We employ the conditional variational autoencoder to make reconstructions of the document, during which the user and product information can be integrated. In the classification module, we build a hierarchical model based on Transformer encoders, where the local-level and global-level prior distribution representations induced from the variational module are injected into the word-level and sentence-level Transformers, respectively. Experimental results on benchmark datasets show that the proposed method significantly outperforms strong baselines, achieving the state-of-the-art performance. Further analysis shows that our model is capable of capturing the latent fine-grained target and opinion prior information, which is highly effective for improving the task performance.

Biomedical information extraction (BioIE) is an important task. The aim is to analyze biomedical texts and extract structured information such as named entities and semantic relations between them. In recent years, pre-trained language models have largely improved the performance of BioIE. However, they neglect to incorporate external structural knowledge, which can provide rich factual information to support the underlying understanding and reasoning for biomedical information extraction. In this paper, we first evaluate current extraction methods, including vanilla neural networks, general language models and pre-trained contextualized language models on biomedical information extraction tasks, including named entity recognition, relation extraction and event extraction. We then propose to enrich a contextualized language model by integrating a large scale of biomedical knowledge graphs (namely, BioKGLM). In order to effectively encode knowledge, we explore a three-stage training procedure and introduce different fusion strategies to facilitate knowledge injection. Experimental results on multiple tasks show that BioKGLM consistently outperforms state-of-the-art extraction models. A further analysis proves that BioKGLM can capture the underlying relations between biomedical knowledge concepts, which are crucial for BioIE.

MOTIVATION: Entity relation extraction is one of the fundamental tasks in biomedical text mining, which is usually solved by the models from natural language processing. Compared with traditional pipeline methods, joint methods can avoid the error propagation from entity to relation, giving better performances. However, the existing joint models are built upon sequential scheme, and fail to detect overlapping entity and relation, which are ubiquitous in biomedical texts. The main reason is that sequential models have relatively weaker power in capturing long-range dependencies, which results in lower performance in encoding longer sentences. In this article, we propose a novel span-graph neural model for jointly extracting overlapping entity relation in biomedical texts. Our model treats the task as relation triplets prediction, and builds the entity-graph by enumerating possible candidate entity spans. The proposed model captures the relationship between the correlated entities via a span scorer and a relation scorer, respectively, and finally outputs all valid relational triplets. RESULTS: Experimental results on two biomedical entity relation extraction tasks, including drug-drug interaction detection and protein-protein interaction detection, show that the proposed method outperforms previous models by a substantial margin, demonstrating the effectiveness of span-graph-based method for overlapping relation extraction in biomedical texts. Further in-depth analysis proves that our model is more effective in capturing the long-range dependencies for relation extraction compared with the sequential models. AVAILABILITY AND IMPLEMENTATION: Related codes are made publicly available at http://github.com/Baxelyne/SpanBioER.

Semantic role labeling (SRL) is one of the important tasks in natural language processing. Current end-to-end SRL, compared with traditional pipeline SRL, has achieved competitive performance via graph-based neural models. However, these are all first-order methods, where decisions for detecting predicate-argument pairs are made in isolation with local short-term features, thus being error-prone. Besides, they may potentially suffer from long-distance dependency, and perform poorly on the data-scarce scenario. In this paper, we explore a second-order end-to-end SRL model that considers simultaneously two pairs of predicate-argument when making scoring. We next propose a structural refinement mechanism to further model higher-order interactions at a global scope. After iterations of refinement, the underlying token representations are refined to be more informative. Experimental results show that our model significantly outperforms the state-of-the-art baselines on multiple benchmarks (i.e., CoNLL08, CoNLL09 and CoNLL12). Further analysis demonstrates that our framework is able to learn latent SRL constraints, being more robust on data-scarce settings, meanwhile relieving the long-range dependency issue.

Pair-wise aspect and opinion terms extraction (PAOTE), aiming at detecting the pair of the correlated aspect and opinion terms jointly, recently has drawn increasing research attention in the community of sentiment analysis and opinion mining. Recent works largely employ joint methods for the task, while they do not sufficiently incorporate the external syntactic knowledge, such as the dependency edges and labels. Besides, these methods fail to capture the underlying shared interactions among the overlapping aspect-opinion pair structures. In this paper we address the above issues for better PAOTE. Specifically, we present a span graph-based model for joint PAOTE, based on which we propose a edge-enhanced syntactic graph convolutional network (ESGCN) to simultaneously encode the syntactic dependency edges and the corresponding labels, for enhancing the extraction and pairing of aspect and opinion terms. During relational pairing, we employ a Biaffine and a Triaffine scorer respectively for high-order scoring, fully exploring the mutual interactions among different aspect-opinion pairs. Experimental results on four benchmark datasets show that our proposed method significantly outperforms all the baseline systems, giving the state-of-the-art performances. Further analysis proves the effectiveness of our framework on modeling the syntactic edge and label knowledge, as well as the advance of the high-order pairing mechanism for the task.

Identifying multiple emotions in a sentence is an important research topic. Existing methods usually model the problem as multi-label classification task. However, previous methods have two issues, limiting the performance of the task. First, these models do not consider prior emotion distribution in a sentence. Second, they fail to effectively capture the context information closely related to the corresponding emotion. In this paper, we propose a Latent Emotion Memory network (LEM) for multi-label emotion classification. The proposed model can learn the latent emotion distribution without external knowledge, and can effectively leverage it into the classification network. Experimental results on two benchmark datasets show that the proposed model outperforms strong baselines, achieving the state-of-the-art performance.

Many efforts of research are devoted to semantic role labeling (SRL) which is crucial for natural language understanding.Supervised approaches have achieved impressing performances when large-scale corpora are available for resource-rich languages such as English.While for the low-resource languages with no annotated SRL dataset, it is still challenging to obtain competitive performances.Cross-lingual SRL is one promising way to address the problem, which has achieved great advances with the help of model transferring and annotation projection.In this paper, we propose a novel alternative based on corpus translation, constructing high-quality training datasets for the target languages from the source gold-standard SRL annotations.Experimental results on Universal Proposition Bank show that the translation-based method is highly effective, and the automatic pseudo datasets can improve the target-language SRL performances significantly.

In this paper, we target improving the joint dialogue act recognition (DAR) and sentiment classification (SC) tasks by fully modeling the local contexts of utterances.First, we employ the dynamic convolution network (DCN) as the utterance encoder to capture the dialogue contexts.Further, we propose a novel context-aware dynamic convolution network (CDCN) to better leverage the local contexts when dynamically generating kernels.We extended our frameworks into bi-channel version (i.e., BDCN and BCDCN) under multi-task learning to achieve the joint DAR and SC.Two channels can learn their own feature representations for DAR and SC, respectively, but with latent interaction.Besides, we suggest enhancing the tasks by employing the DiaBERT language model.Our frameworks 1 obtain state-of-the-art performances against all baselines on two benchmark datasets, demonstrating the importance of modeling the local contexts.

We consider retrofitting structure-aware Transformer language model for facilitating end tasks by proposing to exploit syntactic distance to encode both the phrasal constituency and dependency connection into the language model.A middle-layer structural learning strategy is leveraged for structure integration, accomplished with main semantic task training under multi-task learning scheme.Experimental results show that the retrofitted structure-aware Transformer language model achieves improved perplexity, meanwhile inducing accurate syntactic phrases.By performing structure-aware fine-tuning, our model achieves significant improvements for both semantic-and syntactic-dependent tasks.

Identifying multiple emotions in a piece of text is an important research topic in the NLP community.Existing methods usually model the task as a multi-label classification problem, while these work has two issues. First, these methods fail to leverage the topic information of the text, which has been shown to be effective for sentiment analysis task. Second, different parts of the text can contribute differently to predicting different emotion labels, so the proposed model needs to capture effective features for each corresponding emotion, which is not considered by existing models. To tackle these problems, we propose a topic-enhanced capsule network, which contains two main parts: a variational autoencoder and a capsule module, for multi-label emotion detection task. Specifically, the variational autoencoder can learn the latent topic information of the text, and the capsule module can capture rich features for corresponding emotion. Experimental results on two benchmark datasets show that the proposed model achieves the current best performance, outperforming previous methods and strong baselines by a large margin.

Prior studies show that cross-lingual semantic role labeling (SRL) can be achieved by model transfer under the help of universal features. In this article, we fill the gap of cross-lingual SRL by proposing an end-to-end SRL model that incorporates a variety of universal features and transfer methods. We study both the bilingual transfer and multi-source transfer, under gold or machine-generated syntactic inputs, pre-trained high-order abstract features, and contextualized multilingual word representations. Experimental results on the Universal Proposition Bank corpus indicate that performances of the cross-lingual SRL can vary by leveraging different cross-lingual features. In addition, whether the features are gold-standard also has an impact on performances. Precisely, we find that gold syntax features are much more crucial for cross-lingual SRL, compared with the automatically-generated ones. Moreover, universal dependency structure features are able to give the best help, and both pre-trained high-order features and contextualized word representations can further bring significant improvements.

Many named entities usually contain nested entities in biomedical texts. Nested entities pose challenge to the task of named entity recognition. Traditional methods try to solve the problem as a graph-structure prediction problem. However, these methods fail to sufficiently capture the boundaries information between nested entities, which limits the performance of the task. In this paper, we take a different view by solving each unique entity type as a separate task, using multi-task learning with dispatched attention to facilitate information exchange between tasks. Results on GENIA corpus show that the proposed method is highly effective, obtaining the best results in the literature.

Drug-drug interaction (DDI) extraction is one of important tasks in biomedical text mining. Compared with traditional pipeline extraction, joint methods enjoy lower error propagation, achieving better performances. However, biomedical texts contain a large number of overlapped relations, and existing joint extraction models fail to capture these entity relations. In this paper, we propose a novel joint model for DDI extraction. Different from the existing sequential models, our span-based neural model treats the task as a relation triplets prediction problem, and builds the entity-graph by enumerating all possible candidate spans, and simultaneously models the relation between the correlated entities. Experimental results show that the proposed method significantly outperforms the strong baselines and previous methods, achieving the state-of-the-art performances.