Large Language Models (LLMs) increasingly leverage Federated Learning (FL) to utilize private, task-specific datasets for fine-tuning while preserving data privacy. However, while federated LLM frameworks effectively enable collaborative training without raw data sharing, they critically lack built-in mechanisms for regulatory compliance like GDPR’s right to be forgotten. Integrating private data heightens concerns over data quality and long-term governance, yet existing distributed training frameworks offer no principled way to selectively remove specific client contributions post-training. Due to distributed data silos, stringent privacy constraints, and the intricacies of interdependent model aggregation, federated LLM unlearning is significantly more complex than centralized LLM unlearning. To address this gap, we introduce Oblivionis, a lightweight learning and unlearning framework that enables clients to selectively remove specific private data during federated LLM training, enhancing trustworthiness and regulatory compliance. By unifying FL and unlearning as a dual optimization objective, we incorporate 6 FL and 5 unlearning algorithms for comprehensive evaluation and comparative analysis, establishing a robust pipeline for federated LLM unlearning. Extensive experiments demonstrate that Oblivionis outperforms local training, achieving a robust balance between forgetting efficacy and model utility, with cross-algorithm comparisons providing clear directions for future LLM development.

Haolei Xu, Haiwen Hong, Hongxing Li, Rui Zhou, Yang Zhang, Longtao Huang, Hui Xue, Yongliang Shen, Weiming Lu, Yueting Zhuang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Ziwen Xu, Chenyan WU, Hengyu Sun, Haiwen Hong, Mengru Wang, Yunzhi Yao, Longtao Huang, Hui Xue, Shumin Deng, Zhixuan Chu, Huajun Chen, Ningyu Zhang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Ziwen Xu, Kewei Xu, Haoming Xu, Haiwen Hong, Longtao Huang, Hui Xue, Ningyu Zhang, Yongliang Shen, Guozhou Zheng, Huajun Chen, Shumin Deng. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Federated domain generalization (FDG) for time-series classification (TSC) poses a critical challenge for modern intelligent web services, which rely on edge-collected time-series signals from diverse mobile applications and web devices (e.g., wearables sensors) to support decision-making. The source heterogeneity and temporal dynamics give rise to out-of-distribution (OOD) patterns, which hinder the model's ability to generalize to previously unseen users and devices. In this work, we propose Federated Generalization via Diversity Generation (FedDiG), a diffusion-based FDG framework that captures intra-client distribution shifts from a frequency-domain perspective and employs cross-frequency sampling to synthesize time-series data with diverse spectral patterns. Specifically, FedDiG first performs frequency-proxy representation learning on clients to serve as diffusion conditions. The server then aggregates client-side frequency proxies to construct a global proxy pool and applies class-wise mixup to create novel frequency features. These features guide a global diffusion model to produce diverse data, enabling the simulation of previously unseen patterns and thereby enhancing model training. Extensive experiments on four cross-domain time-series benchmarks demonstrate that FedDiG significantly outperforms state-of-the-art federated learning and FDG baselines, particularly under small-data regimes and large-scale client scenarios, achieving robust generalization to unseen domains in federated settings. This work bridges distribution-diversity synthesis and FDG for time-series to support robust, scalable web applications fed by edge-collected signals, delivering web-scale generalization across heterogeneous web, mobile, and IoT clients.

Federated learning (FL) has emerged as a promising paradigm for privacy-preserving collaborative machine learning. However, extending FL to class incremental learning settings introduces three key challenges: 1) spatial heterogeneity due to non-IID data distributions across clients, 2) temporal heterogeneity due to sequential arrival of tasks, and 3) resource heterogeneity due to diverse client capabilities. Existing approaches generally address these challenges in isolation, potentially leading to interference between updates, catastrophic forgetting or excessive communication overhead. In this paper, we propose personalized Federated class-incremental parameter efficient fine-tuning with Mixture of Frequency aggregation (pFedMxF), a novel framework that simultaneously addresses all three heterogeneity challenges through frequency domain decomposition. Our key insight is that assigning orthogonal frequency components to different clients and tasks enables interference-free learning to be achieved with minimal communication costs. We further design an Auto-task Agnostic Classifier that automatically routes samples to task-specific classifiers while adapting to heterogeneous class distributions. We conduct extensive experiments on three benchmark datasets, comparing our approach with eight state-of-the-art methods. The results demonstrate that pFedMxF achieves comparable test accuracy, while requiring less model parameters and incurring significantly lower communication costs than baseline methods.

The continuous emergence of new objects in the visual world poses a serious challenge to deep object recognition methods, which sparks the increasing study on continual or incremental learning. However, learning new tasks faces the tough catastrophic forgetting problem, i.e., dramatic performance degradation on old tasks. A good continual learning model should be robustly adapted to the upcoming tasks while effectively handling catastrophic forgetting. In this paper, we focus on the class-incremental learning (CIL) task, and propose a novel prototype-based continual learning model C-HPN that projects the visual features into a hypersphere geometric space, where continual learning is conducted. C-HPN features two-fold contributions. On the one hand, instead of using the popular cross-entropy loss, we develop an instance-prototype compact loss to obtain well-clustered hyperspherical embeddings and a prototype-prototype separability loss to boost the model’s generalization by introducing large angle distance inductive bias between prototypes in the hyperspherical space. On the other hand, prototype construction and adaptation strategies are designed for effectively adapting new classes, and an instance-prototype relationship preservation distillation mechanism is introduced to overcome catastrophic forgetting. Extensive experiments on several image datasets validate the effectiveness of the proposed method.

Multimodal Large Language Models (Multimodal LLMs) have shown their strength in Video Question Answering (VideoQA). However, due to the black-box nature of end-to-end training strategies, existing approaches based on Multimodal LLMs suffer from the lack of interpretability for VideoQA: they can neither present reasoning paths nor indicate where the answers are derived from the video. To address this issue, we propose **MSR-ViR** (**M**odularized **S**elf-**R**eflected **Vi**deo **R**easoner), which for the first time integrates modular networks to Multimodal LLMs, capable of providing VideoQA with explicit reasoning paths for more interpretability. Specifically, a **MoST-Grounding** (Modularized Spatial-Temporal Grounding) network is proposed to decompose complex questions via tree-structured policies, localizing relevant temporal and spatial segments within videos through step-by-step reasoning. The proposed MoST-Grounding network provides explicit visually grounded information for Multimodal LLMs with clear reasoning paths, thus enhancing interpretability for the predicted answers. To further improve the reasoning quality, we design an **Alternate Self-reflection Training Strategy** to jointly optimize policy generation and Multimodal LLMs. Experiments on real-world datasets demonstrate the superiority of our proposed MSR-ViR framework in video understanding, reasoning transparency, and providing explicit localization evidence for answers.

Continual multimodal instruction tuning is crucial for adapting Multimodal Large Language Models (MLLMs) to evolving tasks. However, most existing methods adopt a fixed architecture, struggling with adapting to new tasks due to static model capacity. We propose to evolve the architecture under parameter budgets for dynamic task adaptation, which remains unexplored and imposes two challenges: 1) task architecture conflict, where different tasks require varying layer-wise adaptations, and 2) modality imbalance, where different tasks rely unevenly on modalities, leading to unbalanced updates. To address these challenges, we propose a novel Dynamic Mixture of Curriculum LoRA Experts (D-MoLE) method, which automatically evolves MLLM's architecture with controlled parameter budgets to continually adapt to new tasks while retaining previously learned knowledge. Specifically, we propose a dynamic layer-wise expert allocator, which automatically allocates LoRA experts across layers to resolve architecture conflicts, and routes instructions layer-wisely to facilitate knowledge sharing among experts. Then, we propose a gradient-based inter-modal continual curriculum, which adjusts the update ratio of each module in MLLM based on the difficulty of each modality within the task to alleviate the modality imbalance problem. Extensive experiments show that D-MoLE significantly outperforms state-of-the-art baselines, achieving a 15 percent average improvement over the best baseline. To the best of our knowledge, this is the first study of continual learning for MLLMs from an architectural perspective.

Multi-label node classification is an important yet under-explored domain in graph mining as many real-world nodes belong to multiple categories rather than just a single one. Although a few efforts have been made by utilizing Graph Convolution Networks (GCNs) to learn node representations and model correlations between multiple labels in the embedding space, they still suffer from the ambiguous feature and ambiguous topology induced by multiple labels, which reduces the credibility of the messages delivered in graphs and overlooks the label correlations on graph data. Therefore, it is crucial to reduce the ambiguity and empower the GCNs for accurate classification. However, this is quite challenging due to the requirement of retaining the distinctiveness of each label while fully harnessing the correlation between labels simultaneously. To address these issues, in this paper, we propose a Correlation-aware Graph Convolutional Network (CorGCN) for multi-label node classification. By introducing a novel Correlation-Aware Graph Decomposition module, CorGCN can learn a graph that contains rich label-correlated information for each label. It then employs a Correlation-Enhanced Graph Convolution to model the relationships between labels during message passing to further bolster the classification process. Extensive experiments on five datasets demonstrate the effectiveness of our proposed CorGCN.

Determining conditional independence (CI) relationships between random variables is a fundamental yet challenging task in machine learning and statistics, especially in high-dimensional settings. Existing generative model-based CI testing methods, such as those utilizing generative adversarial networks (GANs), often struggle with undesirable modeling of conditional distributions and training instability, resulting in subpar performance. To address these issues, we propose a novel CI testing method via score-based generative modeling, which achieves precise Type I error control and strong testing power. Concretely, we first employ a sliced conditional score matching scheme to accurately estimate conditional score and use Langevin dynamics conditional sampling to generate null hypothesis samples, ensuring precise Type I error control. Then, we incorporate a goodness-of-fit stage into the method to verify generated samples and enhance interpretability in practice. We theoretically establish the error bound of conditional distributions modeled by score-based generative models and prove the validity of our CI tests. Extensive experiments on both synthetic and real-world datasets show that our method significantly outperforms existing state-of-the-art methods, providing a promising way to revitalize generative model-based CI testing.

Cross-lingual representation learning transfers knowledge from resource-rich data to resource-scarce ones to improve the semantic understanding abilities of different languages. However, previous works rely on shallow unsupervised data generated by token surface matching, regardless of the global context-aware semantics of the surrounding text tokens. In this paper, we propose an Unsupervised Pseudo Semantic Data Augmentation (UniPSDA) mechanism for cross-lingual natural language understanding to enrich the training data without human interventions. Specifically, to retrieve the tokens with similar meanings for the semantic data augmentation across different languages, we propose a sequential clustering process in 3 stages: within a single language, across multiple languages of a language family, and across languages from multiple language families. Meanwhile, considering the multi-lingual knowledge infusion with context-aware semantics while alleviating computation burden, we directly replace the key constituents of the sentences with the above-learned multi-lingual family knowledge, viewed as pseudo-semantic. The infusion process is further optimized via three de-biasing techniques without introducing any neural parameters. Extensive experiments demonstrate that our model consistently improves the performance on general zero-shot cross-lingual natural language understanding tasks, including sequence classification, information extraction, and question answering.

Knowledge-enhanced pre-trained language models (KEPLMs) leverage relation triples from knowledge graphs (KGs) and integrate these external data sources into language models via self-supervised learning. Previous works treat knowledge enhancement as two independent operations, i.e., knowledge injection and knowledge integration. In this paper, we propose to learn Knowledge-Enhanced language representations with Hierarchical Reinforcement Learning (KEHRL), which jointly addresses the problems of detecting positions for knowledge injection and integrating external knowledge into the model in order to avoid injecting inaccurate or irrelevant knowledge. Specifically, a high-level reinforcement learning (RL) agent utilizes both internal and prior knowledge to iteratively detect essential positions in texts for knowledge injection, which filters out less meaningful entities to avoid diverting the knowledge learning direction. Once the entity positions are selected, a relevant triple filtration module is triggered to perform low-level RL to dynamically refine the triples associated with polysemic entities through binary-valued actions. Experiments validate KEHRL’s effectiveness in probing factual knowledge and enhancing the model’s performance on various natural language understanding tasks.

KEPLMs are pre-trained models that utilize external knowledge to enhance language understanding. Previous language models facilitated knowledge acquisition by incorporating knowledge-related pre-training tasks learned from relation triples in knowledge graphs. However, these models do not prioritize learning embeddings for entity-related tokens. Updating all parameters in KEPLM is computationally demanding. This paper introduces TRELM, a Robust and Efficient Pre-training framework for Knowledge-Enhanced Language Models. We observe that text corpora contain entities that follow a long-tail distribution, where some are suboptimally optimized and hinder the pre-training process. To tackle this, we employ a robust approach to inject knowledge triples and employ a knowledge-augmented memory bank to capture valuable information. Moreover, updating a small subset of neurons in the feed-forward networks (FFNs) that store factual knowledge is both sufficient and efficient. Specifically, we utilize dynamic knowledge routing to identify knowledge paths in FFNs and selectively update parameters during pre-training. Experimental results show that TRELM achieves at least a 50% reduction in pre-training time and outperforms other KEPLMs in knowledge probing tasks and multiple knowledge-aware language understanding tasks.

Retrieval-augmented large language models (LLMs) leverage relevant content retrieved by information retrieval systems to generate correct responses, aiming to alleviate the hallucination problem. However, existing retriever-responder methods typically append relevant documents to the prompt of LLMs to perform text generation tasks without considering the interaction of fine-grained structural semantics between the retrieved documents and the LLMs. This issue is particularly important for accurate response generation as LLMs tend to “lose in the middle” when dealing with input prompts augmented with lengthy documents. In this work, we propose a new pipeline named “Reinforced Retriever-Reorder-Responder” (R4) to learn document orderings for retrieval-augmented LLMs, thereby further enhancing their generation abilities while the large numbers of parameters of LLMs remain frozen. The reordering learning process is divided into two steps according to the quality of the generated responses: document order adjustment and document representation enhancement. Specifically, document order adjustment aims to organize retrieved document orderings into beginning, middle, and end positions based on graph attention learning, which maximizes the reinforced reward of response quality. Document representation enhancement further refines the representations of retrieved documents for responses of poor quality via document-level gradient adversarial learning. Extensive experiments demonstrate that our proposed pipeline achieves better factual question-answering performance on knowledge-intensive tasks compared to strong baselines across various public datasets. The source codes and trained models will be released upon paper acceptance.

Model editing aims to correct outdated or erroneous knowledge in large language models (LLMs) without the need for costly retraining.Lifelong model editing is the most challenging task that caters to the continuous editing requirements of LLMs.Prior works primarily focus on single or batch editing; nevertheless, these methods fall short in lifelong editing scenarios due to catastrophic knowledge forgetting and the degradation of model performance.Although retrieval-based methods alleviate these issues, they are impeded by slow and cumbersome processes of integrating the retrieved knowledge into the model.In this work, we introduce RECIPE, a RetriEval-augmented ContInuous Prompt lEarning method, to boost editing efficacy and inference efficiency in lifelong learning.RECIPE first converts knowledge statements into short and informative continuous prompts, prefixed to the LLM's input query embedding, to efficiently refine the response grounded on the knowledge.It further integrates the Knowledge Sentinel (KS) that acts as an intermediary to calculate a dynamic threshold, determining whether the retrieval repository contains relevant knowledge.Our retriever and prompt encoder are jointly trained to achieve editing properties, i.e., reliability, generality, and locality.In our experiments, RECIPE is assessed extensively across multiple LLMs and editing datasets, where it achieves superior editing performance.RECIPE also demonstrates its capability to maintain the overall performance of LLMs alongside showcasing fast editing and inference speed.

The social biases and unwelcome stereotypes revealed by pretrained language models are becoming obstacles to their application. Compared to numerous debiasing methods targeting word level, there has been relatively less attention on biases present at phrase level, limiting the performance of debiasing in discipline domains. In this paper, we propose an automatic multi-token debiasing pipeline called General Phrase Debiaser, which is capable of mitigating phrase-level biases in masked language models. Specifically, our method consists of a phrase filter stage that generates stereotypical phrases from Wikipedia pages as well as a model debias stage that can debias models at the multi-token level to tackle bias challenges on phrases. The latter searches for prompts that trigger model’s bias, and then uses them for debiasing. State-of-the-art results on standard datasets and metrics show that our approach can significantly reduce gender biases on both career and multiple disciplines, across models with varying parameter sizes.

Graph Neural Networks (GNNs) exhibit strong potential in node classification task through a message-passing mechanism. However, their performance often hinges on high-quality node labels, which are challenging to obtain in real-world scenarios due to unreliable sources or adversarial attacks. Consequently, label noise is common in real-world graph data, negatively impacting GNNs by propagating incorrect information during training. To address this issue, the study of Graph Neural Networks under Label Noise (GLN) has recently gained traction. However, due to variations in dataset selection, data splitting, and preprocessing techniques, the community currently lacks a comprehensive benchmark, which impedes deeper understanding and further development of GLN. To fill this gap, we introduce NoisyGL in this paper, the first comprehensive benchmark for graph neural networks under label noise. NoisyGL enables fair comparisons and detailed analyses of GLN methods on noisy labeled graph data across various datasets, with unified experimental settings and interface. Our benchmark has uncovered several important insights that were missed in previous research, and we believe these findings will be highly beneficial for future studies. We hope our open-source benchmark library will foster further advancements in this field. The code of the benchmark can be found in https://github.com/eaglelab-zju/NoisyGL.

With the evergrowing sizes of pre-trained models (PTMs), it has been an emerging practice to only provide the inference APIs for users, namely model-as-a-service (MaaS) setting.To adapt PTMs with model parameters frozen, most current approaches focus on the input side, seeking for powerful prompts to stimulate models for correct answers.However, we argue that input-side adaptation could be arduous due to the lack of gradient signals and they usually require thousands of API queries, resulting in high computation and time costs.In light of this, we present Decoder Tuning (DecT), which in contrast optimizes task-specific decoder networks on the output side.Specifically, DecT first extracts prompt-stimulated output scores for initial predictions.On top of that, we train an additional decoder network on the output representations to incorporate posterior data knowledge.By gradientbased optimization, DecT can be trained within several seconds and requires only one PTM query per sample.Empirically, we conduct extensive natural language understanding experiments and show that DecT significantly outperforms state-of-the-art algorithms with a 200× speed-up.Our codes are available at https://github.com/thunlp/DecT.

Large Language Models (LLMs) have made remarkable advancements in the field of natural language generation.However, the propensity of LLMs to generate inaccurate or non-factual content, termed "hallucinations", remains a significant challenge.Current hallucination detection methods often necessitate the retrieval of great numbers of relevant evidence, thereby increasing response times.We introduce a unique framework that leverages statistical decision theory and Bayesian sequential analysis to optimize the trade-off between costs and benefits during the hallucination detection process.This approach does not require a predetermined number of observations.Instead, the analysis proceeds in a sequential manner, enabling an expeditious decision towards "belief" or "disbelief" through a stop-or-continue strategy.Extensive experiments reveal that this novel framework surpasses existing methods in both efficiency and precision of hallucination detection.Furthermore, it requires fewer retrieval steps on average, thus decreasing response times 1 .

Incorporating knowledge graphs (KGs) into recommender systems to provide explainable recommendation has attracted much attention recently. The multi-hop paths in KGs can provide auxiliary facts for improving recommendation performance as well as explainability. However, existing studies may suffer from two major challenges: error propagation and weak explainability. Considering all paths between every user-item pair might involve irrelevant ones, which leads to error propagation of user preferences. Defining meta-paths might alleviate the error propagation, but the recommendation performance would heavily depend on the pre-defined meta-paths. Some recent methods based on graph convolution network (GCN) achieve better recommendation performance, but fail to provide explainability. To tackle the above problems, we propose a novel method named K nowledge-aware R easoning with G raph C onvolution N etwork (KR-GCN). Specifically, to alleviate the effect of error propagation, we design a transition-based method to determine the triple-level scores and utilize nucleus sampling to select triples within the paths between every user-item pair adaptively. To improve the recommendation performance and guarantee the diversity of explanations, user-item interactions and knowledge graphs are integrated into a heterogeneous graph, which is performed with the graph convolution network. A path-level self-attention mechanism is adopted to discriminate the contributions of different selected paths and predict the interaction probability, which improves the relevance of the final explanation. Extensive experiments conducted on three real-world datasets show that KR-GCN consistently outperforms several state-of-the-art baselines. And human evaluation proves the superiority of KR-GCN on explainability.

Prompt-based tuning for pre-trained language models (PLMs) has shown its effectiveness in few-shot learning.Typically, prompt-based tuning wraps the input text into a cloze question.To make predictions, the model maps the output words to labels via a verbalizer, which is either manually designed or automatically built.However, manual verbalizers heavily depend on domain-specific prior knowledge and human efforts, while finding appropriate label words automatically still remains challenging.In this work, we propose the prototypical verbalizer (ProtoVerb) which is built directly from training data.Specifically, Pro-toVerb learns prototype vectors as verbalizers by contrastive learning.In this way, the prototypes summarize training instances and are able to enclose rich class-level semantics.We conduct experiments on both topic classification and entity typing tasks, and the results demonstrate that ProtoVerb significantly outperforms current automatic verbalizers, especially when training data is extremely scarce.More surprisingly, ProtoVerb consistently boosts promptbased tuning even on untuned PLMs, indicating an elegant non-tuning way to utilize PLMs.Our codes are avaliable at https: //github.com/thunlp/OpenPrompt.

Textual adversarial samples play important roles in multiple subfields of NLP research, including security, evaluation, explainability, and data augmentation.However, most work mixes all these roles, obscuring the problem definitions and research goals of the security role that aims to reveal the practical concerns of NLP models.In this paper, we rethink the research paradigm of textual adversarial samples in security scenarios.We discuss the deficiencies in previous work and propose our suggestions that the research on the Security-oriented adversarial NLP (SoadNLP) should: (1) evaluate their methods on security tasks to demonstrate the real-world concerns; (2) consider realworld attackers' goals, instead of developing impractical methods.To this end, we first collect, process, and release a security datasets collection Advbench.Then, we reformalize the task and adjust the emphasis on different goals in SoadNLP.Next, we propose a simple method based on heuristic rules that can easily fulfill the actual adversarial goals to simulate real-world attack methods.We conduct experiments on both the attack and the defense sides on Advbench.Experimental results show that our method has higher practical value, indicating that the research paradigm in SoadNLP may start from our new benchmark.All the code and data of Advbench can be obtained at https: //github.com/thunlp/Advbench.

Capturing emotions within a conversation plays an essential role in modern dialogue systems.However, the weak correlation between emotions and semantics brings many challenges to emotion recognition in conversation (ERC).Even semantically similar utterances, the emotion may vary drastically depending on contexts or speakers.In this paper, we propose a Supervised Prototypical Contrastive Learning (SPCL) loss for the ERC task.Leveraging the Prototypical Network, the SPCL targets at solving the imbalanced classification problem through contrastive learning and does not require a large batch size.Meanwhile, we design a difficulty measure function based on the distance between classes and introduce curriculum learning to alleviate the impact of extreme samples.We achieve state-of-the-art results on three widely used benchmarks.Further, we conduct analytical experiments to demonstrate the effectiveness of our proposed SPCL and curriculum learning strategy.We release the code at https://github.com/caskcsg/SPCL.⋆

Emotion recognition in conversations (ERC) has attracted increasing interests in recent years, due to its wide range of applications, such as customer service analysis, health-care consultation, etc. One key challenge of ERC is that users' emotions would change due to the impact of others' emotions. That is, the emotions within the conversation can spread among the communication participants. However, the spread impact of emotions in a conversation is rarely addressed in existing researches. To this end, we propose EmotionFlow for ERC with the consideration of the spread of participants' emotions during a conversation. EmotionFlow first encodes users' utterance by concatenating the context with an auxiliary question, which helps to learn user-specific features. Then, conditional random field is applied to capture the sequential information at emotional level. We conduct extensive experiments on a public dataset Multimodal EmotionLines Dataset (MELD), and the results demonstrate the effectiveness of our proposed model.

Virtual try-on (VTON) aims at fitting target clothes to reference person images, which is widely adopted in e-commerce. Existing VTON approaches can be narrowly categorized into Parser-Based (PB) and Parser-Free (PF) by whether relying on the parser information to mask the persons’clothes and synthesize try-on images. Although abandoning parser information has improved the applicability of PF methods, the ability of detail synthesizing has also been sacrificed. As a result, the distraction from original cloth may persist in synthesized images, especially in complicated postures and high resolution applications. To address the aforementioned issue, we propose a novel PF method named Regional Mask Guided Network (RMGN). More specifically, a regional mask is proposed to explicitly fuse the features of target clothes and reference persons so that the persisted distraction can be eliminated. A posture awareness loss and a multi-level feature extractor are further proposed to handle the complicated postures and synthesize high resolution images. Extensive experiments demonstrate that our proposed RMGN outperforms both state-of-the-art PB and PF methods. Ablation studies further verify the effectiveness of modules in RMGN. Code is available at https://github.com/jokerlc/RMGN-VITON.

Aspect terms extraction and opinion terms extraction are two key problems of fine-grained Aspect Based Sentiment Analysis (ABSA).The aspect-opinion pairs can provide a global profile about a product or service for consumers and opinion mining systems.However, traditional methods can not directly output aspect-opinion pairs without given aspect terms or opinion terms.Although some recent co-extraction methods have been proposed to extract both terms jointly, they fail to extract them as pairs.To this end, this paper proposes an end-to-end method to solve the task of Pair-wise Aspect and Opinion Terms Extraction (PAOTE).Furthermore, this paper treats the problem from a perspective of joint term and relation extraction rather than under the sequence tagging formulation performed in most prior works.We propose a multi-task learning framework based on shared spans, where the terms are extracted under the supervision of span boundaries.Meanwhile, the pair-wise relations are jointly identified using the span representations.Extensive experiments show that our model consistently outperforms stateof-the-art methods.

As the emergence of E-commerce services, billions of products are sold online everyday. How to detect illegal products from the large-scale online products has become an important and practical research problem. In order to evade detection, malicious sellers usually utilize camouflaged text to describe their illegal products implicitly. Thus brings great challenges to the current detection systems since newly camouflaged text can hardly be learned from historical data and the distribution of illegal and normal products is extremely unbalanced. Rather than solving this problem as a classification task in most previous efforts, we reformulate the problem from a perspective of implicit entity linking, which targets at linking a camouflaged description to a known product. In this paper, we introduce three types of context that could help to infer implicit entity from camouflaged descriptions and propose an end-to-end contextual representation model to capture the effect of different context. Furthermore, we involve a symmetric metric to model the matching score of the input title to the product by learning the mutual effect among the context. The experimental results on the datasets collected from a real-world E-commerce site demonstrate the advantage of the proposed model against the state-of-the-art methods.

Scripts represent knowledge of event sequences that can help text understanding. Script event prediction requires to measure the relation between an existing chain and the subsequent event. The dominant approaches either focus on the effects of individual events, or the influence of the chain sequence. However, only considering individual events will lose much semantic relations within the event chain, and only considering the sequence of the chain will introduce much noise. With our observations, both the individual events and the event segments within the chain can facilitate the prediction of the subsequent event. This paper develops self attention mechanism to focus on diverse event segments within the chain and the event chain is represented as a set of event segments. We utilize the event-level attention to model the relations between subsequent events and individual events. Then, we propose the chain-level attention to model the relations between subsequent events and event segments within the chain. Finally, we integrate event-level and chain-level attentions to interact with the chain to predict what happens next. Comprehensive experiment results on the widely used New York Times corpus demonstrate that our model achieves better results than other state-of-the-art baselines by adopting the evaluation of Multi-Choice Narrative Cloze task.

Target-Based Sentiment Analysis aims at extracting opinion targets and classifying the sentiment polarities expressed on each target. Recently, token based sequence tagging methods have been successfully applied to jointly solve the two tasks, which aims to predict a tag for each token. Since they do not treat a target containing several words as a whole, it might be difficult to make use of the global information to identify that opinion target, leading to incorrect extraction. Independently predicting the sentiment for each token may also lead to sentiment inconsistency for different words in an opinion target. In this paper, inspired by span-based methods in NLP, we propose a simple and effective joint model to conduct extraction and classification at span level rather than token level. Our model first emulates spans with one or more tokens and learns their representation based on the tokens inside. And then, a span-aware attention mechanism is designed to compute the sentiment information towards each span. Extensive experiments on three benchmark datasets show that our model consistently outperforms the state-of-the-art methods.

Obfuscated language is created to avoid censorship in adversarial communication such as sensitive information conveying, strong sentiment expression, secret actions plan, and illegal trading. The obfuscated sentences are usually generated by replacing one word with another to conceal the textual content. Intelligence and security agencies identify such adversarial messages by scanning with a watch-list of red-flagged terms. Though semantic expansion techniques are adopted, the precision and recall of the identification is limited due to the ambiguity and the unbounded creation way. To this end, this paper frames the obfuscated language identification problem as a text matching task, where each message is checked whether matches a red-flagged term. We propose a multimodal text matching model which combining textual and visual features. The proposed model extends a Bi-directional Long Short Term Memory network with a visual-level representation component to achieve the given task. Comparative experiments on real-world dataset demonstrate that the proposed method could achieve a better performance than the previous methods.

An entity on the web can be referred by numerous morphs that are always ambiguous, implicit and informal, which makes it challenging to accurately identify all the morphs corresponding to a specific entity. In this paper, we introduce a novel method based on knowledge graph, which takes advantage of both knowledge reasoning and statistic learning. First, we present a model to build a knowledge graph for the given entity. The knowledge graph integrates the fragmented knowledge on how humans create morphs. Then, the candidate morphs are generated based on the rules summarized from the knowledge graph. At last, we use a classification method to filter the useless candidates and identify the target morphs. The experiments conducted on real world dataset demonstrate efficiency of our proposed method in terms of precision and recall.

With the emergence of online social networks, the social network-based recommendation approach is popularly used. The major benefit of this approach is the ability of dealing with the problems with cold-start users. In addition to social networks, user trust information also plays an important role to obtain reliable recommendations. Although matrix factorization (MF) becomes dominant in recommender systems, the recommendation largely relies on the initialization of the user and item latent feature vectors. Aiming at addressing these challenges, we develop a novel trust-based approach for recommendation in social networks. In particular, we attempt to leverage deep learning to determinate the initialization in MF for trust-aware social recommendations and to differentiate the community effect in user's trusted friendships. A two-phase recommendation process is proposed to utilize deep learning in initialization and to synthesize the users' interests and their trusted friends' interests together with the impact of community effect for recommendations. We perform extensive experiments on real-world social network data to demonstrate the accuracy and effectiveness of our proposed approach in comparison with other state-of-the-art methods.

The advances in mobile technologies enable mobile devices to perform tasks that are traditionally run by personal computers as well as provide services to the others. Mobile users can form a service sharing community within an area by using their mobile devices. This paper highlights several challenges involved in building such service compositions in mobile communities when both service requesters and providers are mobile. To deal with them, we first propose a mobile service provisioning architecture named a mobile service sharing community and then propose a service composition approach by utilizing the Krill-Herd algorithm. To evaluate the effectiveness and efficiency of our approach, we build a simulation tool. The experimental results demonstrate that our approach can obtain superior solutions as compared with current standard composition methods in mobile environments. It can yield near-optimal solutions and has a nearly linear complexity with respect to a problem size.

The development of mobile computing and cloud computing enables people to invoke various services from their mobile devices. This paper focuses on the problem of service composition with temporal and QoS constraints in mobile cloud computing. This problem indeed becomes dramatically tough due to the mobility environment, local-time constraints and global QoS constraints as well. This study aims to form such a service composition that not only satisfies both the time constraints and QoS constraints in a mobile service composition, but also ensures the composition to be executed successfully to the greatest extent in the uncertain mobile environment. Firstly, it formally defines the problem and then transfers the problem into a constrained optimization problem and proves it to be an NP-hard problem. To solve this problem, it proposes a differential evolutionary for constraint driven service composition algorithm (DE4CDSC). It first utilizes a constraints based service filtering process to reduce the searching space and then adopts a differential evolutionary based algorithm to form a service combination by maximizing its successful probability after considering service providers' mobility. To evaluate the proposed approach, a series of simulation experiments and comparisons are conducted to demonstrate the effectiveness of the proposed approach.

The advances in mobile technologies enable us to consume or even provide services through powerful mobile devices anytime and anywhere. Services running on mobile devices within limited range can be composed to coordinate together through wireless communication technologies and perform complex tasks. However, the mobility of users and devices in mobile environment imposes high risk on the execution of the tasks. This paper targets reducing this risk by constructing a dependable service composition after considering the mobility of both service requesters and providers. It first proposes a risk model and clarifies the risk of mobile service composition; and then proposes a service composition approach by modifying the simulated annealing algorithm. Our objective is to form a service composition by selecting mobile services under the mobility model and to ensure the service composition have the best quality of service and the lowest risk. The experimental results demonstrate that our approach can yield near-optimal solutions and has a nearly linear complexity with respect to a problem size.

Mobile business is becoming a reality due to ubiquitous Internet connectivity, popular mobile devices, and widely available cloud services. However, characteristics of the mobile environment, such as mobility, unpredictability, and variation of mobile network's signal strength, present challenges in selecting optimal services for composition. Traditional QoS-aware methods that select individual services with the best QoS may not always result in the best composite service because constant mobility makes the performance of service invocation unpredictable and location-based. This paper discusses the challenges of this problem and defines it in a formal way. To solve this new research problem, we propose a mobility model, a mobility-aware QoS computation rule, and a mobility-enabled selection algorithm with teaching-learning-based optimization. The experimental simulation results demonstrate that our approach can obtain better solutions than current standard composition methods in mobile environments. The approach can obtain near-optimal solutions and has a nearly linear algorithmic complexity with respect to the problem size.

The development of cloud computing and virtualization techniques enables mobile devices to overcome the severity of scarce resource constrained by allowing them to offload computation and migrate several computation parts of an application to powerful cloud servers. A mobile device should judiciously determine whether to offload computation as well as what portion of an application should be offloaded to the cloud. This paper considers a mobile computation offloading problem where multiple mobile services in workflows can be invoked to fulfill their complex requirements and makes decision on whether the services of a workflow should be offloaded. Due to the mobility of portable devices, unstable connectivity of mobile networks can impact the offloading decision. To address this issue, we propose a novel offloading system to design robust offloading decisions for mobile services. Our approach considers the dependency relations among component services and aims to optimize execution time and energy consumption of executing mobile services. To this end, we also introduce a mobility model and a trade-off fault-tolerance mechanism for the offloading system. A genetic algorithm (GA) based offloading method is then designed and implemented after carefully modifying parts of a generic GA to match our special needs for the stated problem. Experimental results are promising and show nearoptimal solutions for all of our studied cases with almost linear algorithmic complexity with respect to the problem size.

Quality-of-Service (QoS)-aware web service composition is of great importance to assemble individual services into a composite one meeting functional and nonfunctional requirements. Given a large number of candidate services, automatic composition is essential so as to derive a composite service efficiently. Most existing methods return one solution that is optimal in some given criteria. This is somewhat rigid in terms of flexibility. In case some component service in the optimal composition becomes unavailable, the composition algorithm has to run again to find another optimal solution. Also, in a lot of circumstances users prefer multiple alternatives over a single one. Therefore, providing top- k service compositions according to their QoS is becoming more desirable. On another aspect, from the perspective of computation efficiency, due to the explosion of the searching space, single-threaded methods are usually not capable of handling a large number of candidate services. This paper tackles these two issues together, i.e., large-scale, QoS-based services composition yielding top- k solutions. The composition algorithm is based on the combination of backtrack search and depth-first search, which can be executed in a parallel way. Experiments are carried out based on the datasets provided by the WS-Challenge competition 2009 and China Web Service 2011. The results show that our approach can not only find the same optimal solution as the winning systems from these competitions, but also provide alternative solutions together with the optimal QoS.

This paper proposes a novel trust evaluation method for service choreography. Compared with current work towards this problem, it considers not only the trust for individual partner services and the explicit trust relation among partner services that have logical dependencies for each other, but also the implicit trust relation implied in data-dependencies among services. A serial of experiments, using the simulation tool Net Logo, are carried out to compare the evaluation results between the proposed method and the method without data-dependency consideration. The result shows that taking consideration of the data-dependency trust improves the accuracy of trust evaluation to a great extent.