Information retrieval (IR) systems play a critical role in navigating information overload across various applications. Existing IR benchmarks primarily focus on simple queries that are semantically analogous to single- and multi-hop relations, overlooking complex logical queries involving first-order logic operations such as conjunction (∧), disjunction (∨), and negation (¬). Thus, these benchmarks can not be used to sufficiently evaluate the performance of IR models on complex queries in real-world scenarios. To address this problem, we propose a novel method leveraging large language models (LLMs) to construct a new IR dataset ComLQ for Complex Logical Queries, which comprises 2,909 queries and 11,251 candidate passages. A key challenge in constructing the dataset lies in capturing the underlying logical structures within unstructured text. Therefore, by designing the subgraph-guided prompt with the subgraph indicator, an LLM (such as GPT-4o) is guided to generate queries with specific logical structures based on selected passages. All query-passage pairs in ComLQ are ensured structure conformity and evidence distribution through expert annotation. To better evaluate whether retrievers can handle queries with negation, we further propose a new evaluation metric, Log-Scaled Negation Consistency (LSNC@K). As a supplement to standard relevance-based metrics (such as nDCG and mAP), LSNC@K measures whether top-K retrieved passages violate negation conditions in queries. Our experimental results under zero-shot settings demonstrate existing retrieval models' limited performance on complex logical queries, especially on queries with negation, exposing their inferior capabilities of modeling exclusion. In summary, our ComLQ offers a comprehensive and fine-grained exploration, paving the way for future research on complex logical queries in IR.

Powei Chang, Jin Xiao, Guanglei Yue, Qianyu He, Yanghua Xiao, Deqing Yang, Jiaqing Liang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Ziheng Li, Liu Kang, Feng Xiao, Luxi Xing, Qingyi Si, Zhuoran Li, Weikang Gong, Deqing Yang, Yanghua Xiao, Hongcheng Guo. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Nianqi Li, Siyu Yuan, Jiangjie Chen, Jiaqing Liang, Feng Wei, Zujie Liang, Deqing Yang, Yanghua Xiao. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Recent advances in large language models (LLMs) have enabled social simulation through multi-agent systems.Prior efforts focus on agent societies created from scratch, assigning agents with newly defined personas.However, simulating established fictional worlds and characters remain largely underexplored, despite its significant practical value.In this paper, we introduce BOOKWORLD, a comprehensive system for constructing and simulating book-based multi-agent societies.BOOK-WORLD's design covers comprehensive realworld intricacies, including diverse and dynamic characters, fictional worldviews, geographical constraints and changes, e.t.c.BOOK-WORLD enables diverse applications including story generation, interactive games and social simulation, offering novel ways to extend and explore beloved fictional works.Through extensive experiments, we demonstrate that BOOKWORLD generates creative, high-quality stories while maintaining fidelity to the source books, surpassing previous methods with a win rate of 75.36%.The code and demo of this paper can be found at the project page: https://bookworld2025.github.io/.

Ruihan Yang, Caiqi Zhang, Zhisong Zhang, Xinting Huang, Sen Yang, Nigel Collier, Dong Yu, Deqing Yang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

User recommendation systems enhance user engagement by encouraging users to act as inviters to interact with other users (invitees), potentially fostering information propagation. Conventional recommendation methods typically focus on modeling interaction willingness. Influence-Maximization (IM) methods focus on identifying a set of users to maximize the information propagation. However, existing methods face two significant challenges. First, recommendation methods fail to unleash the candidates' spread capability. Second, IM methods fail to account for the willingness to interact. To solve these issues, we propose two models named HeteroIR and HeteroIM. HeteroIR provides an intuitive solution to unleash the dissemination potential of user recommendation systems. HeteroIM fills the gap between the IM method and the recommendation task, improving interaction willingness and maximizing spread coverage. The HeteroIR introduces a two-stage framework to estimate the spread profits. The HeteroIM incrementally selects the most influential invitee to recommend and rerank based on the number of reverse reachable (RR) sets containing inviters and invitees. RR set denotes a set of nodes that can reach a target via propagation. Extensive experiments show that HeteroIR and HeteroIM significantly outperform the state-of-the-art baselines with the p-value<0.05. Furthermore, we have deployed HeteroIR and HeteroIM in Tencent's online gaming platforms and gained an 8.5% and 10% improvement in the online A/B test, respectively. Implementation codes are available at https://github.com/socialalgo/HIM.

Many previous models of named entity recognition (NER) suffer from the problem of Out-of-Entity (OOE), i.e., the tokens in the entity mentions of the test samples have not appeared in the training samples, which hinders the achievement of satisfactory performance. To improve OOE-NER performance, in this paper, we propose a new framework, namely S+NER, which fully leverages sentence-level information. Our S+NER achieves better OOE-NER performance mainly due to the following two particular designs. 1) It first exploits the pre-trained language model’s capability of understanding the target entity’s sentence-level context with a template set. 2) Then, it refines the sentence-level representation based on the positive and negative templates, through a contrastive learning strategy and template pooling method, to obtain better NER results. Our extensive experiments on five benchmark datasets have demonstrated that, our S+NER outperforms some state-of-the-art OOE-NER models.

The task of translating natural language questions into query languages has long been a central focus in semantic parsing.Recent advancements in Large Language Models (LLMs) have significantly accelerated progress in this field.However, existing studies typically focus on a single query language, resulting in methods with limited generalizability across different languages.In this paper, we formally define the Text-to-Query task paradigm, unifying semantic parsing tasks across various query languages.We identify query skeletons as a shared optimization target of Text-to-Query tasks, and propose a general dynamic data augmentation framework that explicitly diagnoses modelspecific weaknesses in handling these skeletons to synthesize targeted training data.Experiments on four Text-to-Query benchmarks demonstrate that our method achieves state-ofthe-art performance using only a small amount of synthesized data, highlighting the efficiency and generality of our approach and laying a solid foundation for unified research on Textto-Query tasks.We release our code

Pun memes, which combine wordplay with visual elements, represent a popular form of humor in Chinese online communications.Despite their prevalence, current Vision-Language Models (VLMs) lack systematic evaluation in understanding and applying these culturallyspecific multimodal expressions.In this paper, we introduce PUNMEMECN, a novel benchmark designed to assess VLMs' capabilities in processing Chinese pun memes across three progressive tasks: pun meme detection, pun meme sentiment analysis, and chat-driven meme response.PUNMEMECN consists of 1,959 Chinese memes (653 pun memes and 1,306 non-pun memes) with comprehensive annotations of punchlines, sentiments, and explanations, alongside 2,008 multi-turn chat conversations incorporating these memes.Our experiments indicate that state-of-the-art VLMs struggle with Chinese pun memes, particularly with homophone wordplay, even with Chainof-Thought prompting.Notably, punchlines in memes can effectively conceal potentially harmful content from AI detection.These findings underscore the challenges in cross-cultural multimodal understanding and highlight the need for culture-specific approaches to humor comprehension in AI systems.

Large Language Models (LLMs) are prone to hallucination, particularly in long-form generations.A promising direction to mitigate hallucination is to teach LLMs to express uncertainty explicitly when they lack sufficient knowledge.However, existing work lacks direct and fair evaluation of LLMs' ability to express uncertainty effectively in long-form generation.To address this gap, we first introduce UNCLE, a benchmark designed to evaluate uncertainty expression in both long-and short-form question answering (QA).UNCLE covers five domains and includes more than 1,000 entities, each with paired short-and long-form QA items.Our dataset is the first to directly link short-and long-form QA through aligned questions and gold-standard answers.Along with UNCLE, we propose a suite of new metrics to assess the models' capabilities to selectively express uncertainty.We then demonstrate that current models fail to convey uncertainty appropriately in long-form generation.We further explore both prompt-based and training-based methods to improve models' performance, with the training-based methods yielding greater gains.Further analysis of alignment gaps between short-and long-form uncertainty expression highlights promising directions for future research using UNCLE.

Existing Large Language Models (LLMs) generate texts through unidirectional autoregressive decoding methods to respond to various queries. These methods tend to consider token selection in a simple sequential manner, making it easy to fall into suboptimal options when encountering uncertain tokens, referred to as chaotic points in this paper. Many chaotic points exist in the texts generated by LLMs, and they often significantly affect the quality of subsequently generated tokens, which would interfere with LLMs’ generation. In this paper, we propose DEval, a decoding framework for enhancing LLMs’ generation. Analogous to human decision-making, DEval integrates speculation and evaluation into decoding, allowing LLMs to make more careful decisions, and thus optimize token selection at chaotic points. Experiments across various tasks using different LLMs demonstrate DEval’s effectiveness. The code and more technical details are available at our repository1.

Ruihan Yang, Jiangjie Chen, Yikai Zhang, Siyu Yuan, Aili Chen, Kyle Richardson, Yanghua Xiao, Deqing Yang. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Siyu Yuan, Kaitao Song, Jiangjie Chen, Xu Tan, Dongsheng Li, Deqing Yang. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Siyu Yuan, Kaitao Song, Jiangjie Chen, Xu Tan, Yongliang Shen, Kan Ren, Dongsheng Li, Deqing Yang. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

This work proposes a novel framework based on nested evolving set processes to accelerate Personalized PageRank (PPR) computation. At each stage of the process, we employ a localized inexact proximal point iteration to solve a simplified linear system. We show that the time complexity of such localized methods is upper bounded by $\min\{\tilde{\mathcal{O}}(R^2/\epsilon^2), \tilde{\mathcal{O}}(m)\}$ to obtain an $\epsilon$-approximation of the PPR vector, where $m$ denotes the number of edges in the graph and $R$ is a constant defined via nested evolving set processes. Furthermore, the algorithms induced by our framework require solving only $\tilde{\mathcal{O}}(1/\sqrt{\alpha})$ such linear systems, where $\alpha$ is the damping factor. When $1/\epsilon^2\ll m$, this implies the existence of an algorithm that computes an $\epsilon$-approximation of the PPR vector with an overall time complexity of $\tilde{\mathcal{O}}(R^2 / (\sqrt{\alpha}\epsilon^2))$, independent of the underlying graph size. Our result resolves an open conjecture from existing literature. Experimental results on real-world graphs validate the efficiency of our methods, demonstrating significant convergence in the early stages.

Large language models (LLMs) have recently transformed from text-based assistants to autonomous agents capable of planning, reasoning, and iteratively improving their actions. While numerical reward signals and verifiers can effectively rank candidate actions, they often provide limited contextual guidance. In contrast, natural language feedback better aligns with the generative capabilities of LLMs, providing richer and more actionable suggestions. However, parsing and implementing this feedback effectively can be challenging for LLM-based agents. In this work, we introduce Critique-Guided Improvement (CGI), a novel two-player framework, comprising an actor model that explores an environment and a critic model that generates detailed nature language feedback. By training the critic to produce fine-grained assessments and actionable revisions, and the actor to utilize these critiques, our approach promotes more robust exploration of alternative strategies while avoiding local optima. Experiments in three interactive environments show that CGI outperforms existing baselines by a substantial margin. Notably, even a small critic model surpasses GPT-4 in feedback quality. The resulting actor achieves state-of-the-art performance, demonstrating the power of explicit iterative guidance to enhance decision-making in LLM-based agents.

Large language models (LLMs) have enabled agents to perform complex reasoning and decision-making through free-form language interactions. However, in open-ended language action environments (e.g., negotiation or question-asking games), the action space can be formulated as a joint distribution over tokens, resulting in an extremely large and combinatorial action space. Sampling actions in such a space can lead to extreme reward sparsity, which brings large reward variance, hindering effective reinforcement learning (RL). To address this, we propose **ARIA**, a method that **A**ggregates **R**ewards in **I**ntention space to enable efficient and effective language **A**gents training. ARIA aims to project natural language actions from the high-dimensional joint token distribution space into a low-dimensional intention space, where semantically similar actions are clustered and assigned shared rewards. This intention-aware reward aggregation reduces reward variance by densifying reward signals, fostering efficient and effective policy optimization. Extensive experiments demonstrate that ARIA not only significantly reduces gradient variance, but also delivers substantial performance gains of average 9.95% across four downstream tasks (e.g., negotiation and text-based games), consistently outperforming strong offline and online RL baselines.

In recent years, the research of employing knowledge graphs (KGs) in sequential recommendation (SR) has received a lot of attention, since the side information extracted from KGs, especially the information of the correlations between items, indeed helps the SR models achieve better performance. However, many previous KG-based SR models tend to introduce some noise information when learning item embeddings, or insufficiently fuse item–item correlations into their sequential modeling, thus limiting their performance improvements . In this paper, we propose a D istance- A ware K nowledge-based S equential R ecommendation model ( DAKSR ), which exploits the explicit item–item correlations from KGs to achieve enhanced SR. Specifically, as one critical component in our DAKSR, the distance score matrix (DSM) is first obtained to indicate the correlations between items, and then leveraged in the following three major modules of DAKSR. First, in the Item-Set Embedding layer (ISE) all item embeddings are learned based on DSM, in which the noise information is eliminated effectively. Meanwhile, the Knowledge-Infused Transformer (KIT) incorporates DSM into its attention mechanism to improve the feature extraction. Furthermore, the Knowledge Contrastive Learning module (KCL) also leverages the item–item correlations presented in DSM to generate two credible sequence views, which are used to refine sample representations through a contrastive learning strategy, and thus improve the model’s robustness. Our extensive experiments on three SR benchmarks obviously demonstrate our DAKSR’s superior performance over the state-of-the-art (SOTA) KG-based recommendation models. The implementation of our DAKSR is available at https://github.com/Easonsi/DAKSR for reproducing our experiment results conveniently.

Analogical reasoning is a fundamental cognitive ability of humans.However, current language models (LMs) still struggle to achieve human-like performance in analogical reasoning tasks due to a lack of resources for model training.In this work, we address this gap by proposing ANALOGYKB, a million-scale analogy knowledge base (KB) derived from existing knowledge graphs (KGs).ANALOGYKB identifies two types of analogies from the KGs: 1) analogies of the same relations, which can be directly extracted from the KGs, and 2) analogies of analogous relations, which are identified with a selection and filtering pipeline enabled by large language models (LLMs), followed by minor human efforts for data quality control.Evaluations on a series of datasets of two analogical reasoning tasks (analogy recognition and generation) demonstrate that ANAL-OGYKB successfully enables both smaller LMs and LLMs to gain better analogical reasoning capabilities.Resources of this paper can be found at https://github.com/siyuyuan/ analogykb.

The relation extraction (RE) in complex scenarios faces challenges such as diverse relation types and ambiguous relations between entities within a single sentence, leading to the poor performance of pure "text-in, text-out" language models (LMs). To address these challenges, in this paper, we propose an agent-based RE framework, namely "AgentRE", which fully leverages the potential of large language models (LLMs) including memory, retrieval and reflection, to achieve RE in complex scenarios. Specifically, three major modules are built in AgentRE serving as the tools to help the agent acquire and process various useful information, thereby obtaining improved RE performance. Our extensive experimental results upon two datasets in English and Chinese demonstrate our AgentRE's superior performance, especially in low-resource scenarios. Additionally, the trajectories generated by AgentRE can be refined to construct a high-quality training dataset incorporating different reasoning methods, which can be used to fine-tune smaller models.

Relation triple extraction, which outputs a set of triples from long sentences, plays a vital role in knowledge acquisition. Large language models can accurately extract triples from simple sentences through few-shot learning or fine-tuning when given appropriate instructions. However, they often miss out when extracting from complex sentences. In this paper, we design an evaluation-filtering framework that integrates large language models with small models for relational triple extraction tasks. The framework includes an evaluation model that can extract related entity pairs with high precision. We propose a simple labeling principle and a deep neural network to build the model, embedding the outputs as prompts into the extraction process of the large model. We conduct extensive experiments to demonstrate that the proposed method can assist large language models in obtaining more accurate extraction results, especially from complex sentences containing multiple relational triples. Our evaluation model can also be embedded into traditional extraction models to enhance their extraction precision from complex sentences.

In recent years, the fine-tuned generative models have been proven more powerful than the previous tagging-based or span-based models on named entity recognition (NER) task. It has also been found that the information related to entities, such as entity types, can prompt a model to achieve NER better. However, it is not easy to determine the entity types indeed existing in the given sentence in advance, and inputting too many potential entity types would distract the model inevitably. To exploit entity types’ merit on promoting NER task, in this paper we propose a novel NER framework, namely ToNER based on a generative model. In ToNER, a type matching model is proposed at first to identify the entity types most likely to appear in the sentence. Then, we append a multiple binary classification task to fine-tune the generative model’s encoder, so as to generate the refined representation of the input sentence. Moreover, we add an auxiliary task for the model to discover the entity types which further fine-tunes the model to output more accurate results. Our extensive experiments on some NER benchmarks verify the effectiveness of our proposed strategies in ToNER that are oriented towards entity types’ exploitation.

Previous works of negation understanding mainly focus on negation cue detection and scope resolution, without identifying negation subject which is also significant to the downstream tasks. In this paper, we propose a new negation triplet extraction (NTE) task which aims to extract negation subject along with negation cue and scope. To achieve NTE, we devise a novel Syntax&Semantic-Enhanced Negation Extraction model, namely SSENE, which is built based on a generative pretrained language model (PLM) of Encoder-Decoder architecture with a multi-task learning framework. Specifically, the given sentence’s syntactic dependency tree is incorporated into the PLM’s encoder to discover the correlations between the negation subject, cue and scope. Moreover, the semantic consistency between the sentence and the extracted triplet is ensured by an auxiliary task learning. Furthermore, we have constructed a high-quality Chinese dataset NegComment based on the users’ reviews from the real-world platform of Meituan, upon which our evaluations show that SSENE achieves the best NTE performance compared to the baselines. Our ablation and case studies also demonstrate that incorporating the syntactic information helps the PLM’s recognize the distant dependency between the subject and cue, and the auxiliary task learning is helpful to extract the negation triplets with more semantic consistency. We further demonstrate that SSENE is also competitive on the traditional CDSR task.

As one of the common rhetorical devices, puns play a vital role in linguistic study, including the comprehensive analysis of linguistic humor.Although large language models (LLMs) have been widely explored on various tasks of natural language understanding and generation, their ability to understand puns has not been systematically studied, limiting the utilization of LLMs in creative writing and humor creation.In this paper, we leverage three popular tasks, i.e., pun recognition, pun explanation, and pun generation, to systematically evaluate LLMs' capability of understanding puns.In addition to the evaluation metrics adopted by prior research, we introduce some new evaluation methods and metrics that are better suited to the in-context learning paradigm of LLMs.These new metrics offer a more rigorous assessment of an LLM's capability to understand puns and align more closely with human cognition.Our research findings reveal the "lazy pun generation" pattern and identify the primary challenges in understanding puns with LLMs.The code is available at https://github.com/Zhijun-Xu/PunEval.

Analogical reasoning plays a critical role in human cognition, enabling us to understand new concepts by associating them with familiar ones.Previous research in the AI community has mainly focused on identifying and generating analogies and then examining their quality under human evaluation, which overlooks the practical application of these analogies in real-world settings.Inspired by the human education process, in this paper, we propose to investigate how analogies created by teacher language models (LMs) can assist student LMs in understanding scientific concepts, thereby aligning more closely with practical scenarios.Our results suggest that free-form analogies can indeed aid LMs in understanding concepts.Additionally, analogies generated by student LMs can improve their own performance on scientific question answering, demonstrating their capability to use analogies for self-learning new knowledge.Resources are available at https://github.com/siyuyuan/SCUA.

Large language models (LLMs) have demonstrated impressive performance and spurred numerous AI applications, in which role-playing agents (RPAs) are particularly popular, especially for fictional characters.The prerequisite for these RPAs lies in the capability of LLMs to understand characters from fictional works.Previous efforts have evaluated this capability via basic classification tasks or characteristic imitation, failing to capture the nuanced character understanding with LLMs.In this paper, we propose evaluating LLMs' character understanding capability via the character profiling task, i.e., summarizing character profiles from corresponding materials, a widely adopted yet understudied practice for RPA development.Specifically, we construct the CROSS dataset from literature experts and assess the generated profiles by comparing them with ground truth references and evaluating their applicability in downstream tasks.Our experiments, which cover various summarization methods and LLMs, have yielded promising results.These results strongly validate the character understanding capability of LLMs.Resources are available at https://github. com/Joanna0123/character_profiling.

Efficient computation of graph diffusion equations (GDEs), such as Personalized PageRank, Katz centrality, and the Heat kernel, is crucial for clustering, training neural networks, and many other graph-related problems. Standard iterative methods require accessing the whole graph per iteration, making them time-consuming for large-scale graphs. While existing local solvers approximate diffusion vectors through heuristic local updates, they often operate sequentially and are typically designed for specific diffusion types, limiting their applicability. Given that diffusion vectors are highly localizable, as measured by the participation ratio, this paper introduces a novel framework for approximately solving GDEs using a local diffusion process. This framework reveals the suboptimality of existing local solvers. Furthermore, our approach effectively localizes standard iterative solvers by designing simple and provably sublinear time algorithms. These new local solvers are highly parallelizable, making them well-suited for implementation on GPUs. We demonstrate the effectiveness of our framework in quickly obtaining approximate diffusion vectors, achieving up to a hundred-fold speed improvement, and its applicability to large-scale dynamic graphs. Our framework could also facilitate more efficient local message-passing mechanisms for GNNs.

Given the damping factor $\alpha$ and precision tolerance $\epsilon$, \citet{andersen2006local} introduced Approximate Personalized PageRank (APPR), the \textit{de facto local method} for approximating the PPR vector, with runtime bounded by $\Theta(1/(\alpha\epsilon))$ independent of the graph size. Recently, Fountoulakis \& Yang asked whether faster local algorithms could be developed using $\tilde{\mathcal{O}}(1/(\sqrt{\alpha}\epsilon))$ operations. By noticing that APPR is a local variant of Gauss-Seidel, this paper explores the question of *whether standard iterative solvers can be effectively localized*. We propose to use the *locally evolving set process*, a novel framework to characterize the algorithm locality, and demonstrate that many standard solvers can be effectively localized. Let $\overline{\operatorname{vol}}{ (\mathcal S_t)}$ and $\overline{\gamma_t}$ be the running average of volume and the residual ratio of active nodes $\textstyle \mathcal{S_t}$ during the process. We show $\overline{\operatorname{vol}}{ (\mathcal S_t)}/\overline{\gamma_t} \leq 1/\epsilon$ and prove APPR admits a new runtime bound $\tilde{\mathcal{O}}(\overline{\operatorname{vol}}(\mathcal S_t)/(\alpha\overline{\gamma_t}))$ mirroring the actual performance. Furthermore, when the geometric mean of residual reduction is $\Theta(\sqrt{\alpha})$, then there exists $c \in (0,2)$ such that the local Chebyshev method has runtime $\tilde{\mathcal{O}}(\overline{\operatorname{vol}}(\mathcal{S_t})/(\sqrt{\alpha}(2-c)))$ without the monotonicity assumption. Numerical results confirm the efficiency of this novel framework and show up to a hundredfold speedup over corresponding standard solvers on real-world graphs.

Open information extraction (OpenIE) aims to extract the schema-free triplets in the form of (subject, predicate, object) from a given sentence. Compared with general information extraction (IE), OpenIE poses more challenges for the IE models, especially when multiple complicated triplets exist in a sentence. To extract these complicated triplets more effectively, in this paper we propose a novel generative OpenIE model, namely DualOIE, which achieves a dual task at the same time as extracting some triplets from the sentence, i.e., converting the triplets into the sentence. Such dual task encourages the model to correctly recognize the structure of the given sentence and thus is helpful to extract all potential triplets from the sentence. Specifically, DualOIE extracts the triplets in two steps: 1) first extracting a sequence of all potential predicates, 2) then using the predicate sequence as a prompt to induce the generation of triplets. Our experiments on two benchmarks and our dataset constructed from Meituan demonstrate that DualOIE achieves the best performance among the state-of-the-art baselines. Furthermore, the online A/B test on Meituan platform shows that 0.93% improvement of QV-CTR and 0.56% improvement of UV-CTR have been obtained when the triplets extracted by DualOIE were leveraged in Meituan's search system.

Siyu Yuan, Jiangjie Chen, Ziquan Fu, Xuyang Ge, Soham Shah, Charles Jankowski, Yanghua Xiao, Deqing Yang. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Siyu Yuan, Deqing Yang, Jinxi Liu, Shuyu Tian, Jiaqing Liang, Yanghua Xiao, Rui Xie. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

In recent years, group buying has become one popular kind of online shopping activities, thanks to its larger sales and lower unit price. Unfortunately, seldom research focuses on the recommendations specifically for group buying by now. Although some recommendation models have been proposed for group recommendation, they can not be directly used to achieve the real-world group buying recommendation, due to the essential difference between group recommendation and group buying recommendation. In this paper, we first formalize the task of group buying recommendation into two sub-tasks. Then, based on our insights into the correlations and interactions between the two sub-tasks, we propose a novel recommendation model for group buying, namely MGBR, which is built mainly with a multi-task learning module. To improve recommendation performance further, we devise some collaborative expert networks and adjusted gates in the multi-task learning module, to promote the information interaction between the two sub-tasks. Furthermore, we propose two auxiliary losses corresponding to the two sub-tasks, to refine the representation learning in our model. Our extensive experiments not only demonstrate that the augmented representations learned in our model result in better performance than previous recommendation models, but also justify the impacts of the specially designed components in our model. To reproduce our model’s recommendation results conveniently, we have provided our model’s source code and dataset on https://github.com/DeqingYang/MGBR.

Personalized PageRank Vectors are widely used as fundamental graph-learning tools for detecting anomalous spammers, learning graph embeddings, and training graph neural networks. The well-known local FwdPush algorithm[5] approximates PPVs and has a sublinear rate of O(1 over αε). A recent study [51] found that when high precision is required, FwdPush is similar to the power iteration method, and its run time is pessimistically bounded by O(m over α log 1 over ε). This paper looks closely at calculating PPVs for both directed and undirected graphs. By leveraging the linear invariant property, we show that FwdPush is a variant of Gauss-Seidel and propose a Successive Over-Relaxation based method, FwdPushSOR to speed it up by slightly modifying FwdPush. Additionally, we prove FwdPush has local linear convergence rate O(vol (S) over α log 1 over ε) enjoying advantages of two existing bounds. We also design a new local heuristic push method that reduces the number of operations by 10-50 percent compared to FwdPush. For undirected graphs, we propose two momentum-based acceleration methods that can be expressed as one-line updates and speed up non-acceleration methods by O (1 / √ α). Our experiments on six real-world graph datasets confirm the efficiency of FwdPushSOR and the acceleration methods for directed and undirected graphs, respectively.

POI tagging aims to annotate a point of interest (POI) with some informative tags, which facilitates many services related to POIs, including search, recommendation, and so on. Most of the existing solutions neglect the significance of POI images and seldom fuse the textual and visual features of POIs, resulting in suboptimal tagging performance. In this paper, we propose a novel M ulti-M odal M odel for P OI T agging, namely M3PT, which achieves enhanced POI tagging through fusing the target POI's textual and visual features, and the precise matching between the multi-modal representations. Specifically, we first devise a domain-adaptive image encoder (DIE) to obtain the image embeddings aligned to their gold tags' semantics. Then, in M3PT's text-image fusion module (TIF), the textual and visual representations are fully fused into the POIs' content embeddings for the subsequent matching. In addition, we adopt a contrastive learning strategy to further bridge the gap between the representations of different modalities. To evaluate the tagging models' performance, we have constructed two high-quality POI tagging datasets from the real-world business scenario of Ali Fliggy. Upon the datasets, we conducted the extensive experiments to demonstrate our model's advantage over the baselines of uni-modality and multi-modality, and verify the effectiveness of important components in M3PT, including DIE, TIF and the contrastive learning strategy.

Sequential recommendation (SR) aims to predict a user’s next interacted item given his/her historical interactions. Most existing sequential recommendation systems model user preferences only with item-level representations, where a user’s interaction sequence are often modeled with sequential or graph-based method to infer the user’s sequential interaction pattern. However, since a user’s preference factors may vary over time, the user modeling on item-level could hardly represent the user’s preference precisely and sufficiently, resulting in suboptimal recommendation performance. In addition, the recommendation results based on the item-level user representations lack the interpretability of preference factors. To address these problems, we propose a novel SR model with dual-view user representations in this paper, namely DUVRec, where a user’s preference is learned based on the representations of two distinct views, i.e., item view and factor view . Specifically, the item-view user representation is learned as the previous SR models to encode the user preference of item level, while the factor-view user representation is learned by an coarse-grained graph embedding method to explicitly represent the user in terms of preference factors. As a result, such dual-view user representations are more comprehensive than that in the previous SR models, leading to enhanced SR performance. Furthermore, we design a contrastive learning strategy to achieve mutual complementation between these two views. Our extensive experiments upon three benchmark datasets justify DUVRec’s superior performance over the state-of-the-art SR models, including the advantage of the dual-view contrastive learning. In addition, DUVRec’s capability of providing explanations on recommendation results is also demonstrated through some specific case studies.

Chinese abbreviation prediction is very important for various natural language processing tasks such as query understanding and entity linking, since people tend to use the concise abbreviation rather than the full form (name) to mention an entity. The existing models achieve their predictions through sequence labeling, i.e., the binary classification for each character (token) of the full form. However, they only leverage the semantics of the entity itself, overlooking the label dependencies between the tokens, and the rich information of the entity-related texts. In this paper we proposed a Context-Enhanced Transformer with Abbr-Recover policy, namely CETAR, for Chinese abbreviation prediction. CETAR predicts the abbreviation sequence mainly through an iterative decoding process, of which each round consists of an abbreviation and recovery operation. Our extensive experiments upon both general field and specific domain datasets justify that CETAR outperforms the state-of-the-art baselines including sequence labeling models and sequence generation models. Moreover, we have successfully constructed a Chinese abbreviation dataset from the famous tour website Fliggy, and we also shared it at https://github.com/tolerancecky/abbr-0731. The online A/B test on the Fliggy search system shows that 2.03% of conversion rate improvement has been achieved with the predicted abbreviations.

Many commonsense knowledge graphs (CKGs) still suffer from incompleteness although they have been applied in many natural language processing tasks successfully. Due to the scale and sparsity of CKGs, existing knowledge base completion models are not still competent for CKGs. In this paper, we propose a commonsense knowledge base completion (CKBC) model which learns the structural representations and contextual representations of CKG nodes and relations, respectively by a relational graph attention network and a pre-trained language model. Based on these two types of representations, the scoring decoder in our model achieves a more accurate prediction for a given triple. Our empirical studies on the representative CKG ConceptNet demonstrate our model's superiority over the state-of-the-art CKBC models.

Continual relation extraction (CRE) aims to extract relations towards the continuous and iterative arrival of new data, of which the major challenge is the catastrophic forgetting of old tasks. In order to alleviate this critical problem for enhanced CRE performance, we propose a novel Continual Relation Extraction framework with Contrastive Learning, namely CRECL, which is built with a classification network and a prototypical contrastive network to achieve the incremental-class learning of CRE. Specifically, in the contrastive network a given instance is contrasted with the prototype of each candidate relations stored in the memory module. Such contrastive learning scheme ensures the data distributions of all tasks more distinguishable, so as to alleviate the catastrophic forgetting further. Our experiment results not only demonstrate our CRECL’s advantage over the state-of-the-art baselines on two public datasets, but also verify the effectiveness of CRECL’s contrastive learning on improving performance.

Entity typing aims to assign types to the entity mentions in given texts.The traditional classification-based entity typing paradigm has two unignorable drawbacks: 1) it fails to assign an entity to the types beyond the predefined type set, and 2) it can hardly handle few-shot and zero-shot situations where many long-tail types only have few or even no training instances.To overcome these drawbacks, we propose a novel generative entity typing (GET) paradigm: given a text with an entity mention, the multiple types for the role that the entity plays in the text are generated with a pre-trained language model (PLM).However, PLMs tend to generate coarse-grained types after finetuning upon the entity typing dataset.In addition, only the heterogeneous training data consisting of a small portion of human-annotated data and a large portion of auto-generated but low-quality data are provided for model training.To tackle these problems, we employ curriculum learning (CL) to train our GET model on heterogeneous data, where the curriculum could be self-adjusted with the self-paced learning according to its comprehension of the type granularity and data heterogeneity.Our extensive experiments upon the datasets of different languages and downstream tasks justify the superiority of our GET model over the state-ofthe-art entity typing models.The code has been released on https://github.com/siyuyuan/GET.

Most sequential recommendation (SR) systems employing graph neural networks (GNNs) only model a user's interaction sequence as a flat graph without hierarchy, overlooking diverse factors in the user's preference. Moreover, the timespan between interacted items is not sufficiently utilized by previous models, restricting SR performance gains. To address these problems, we propose a novel SR system employing a hierarchical graph neural network (HGNN) to model factorial user preferences. Specifically, a timespan-aware sequence graph (TSG) for the target user is first constructed with the timespan among interacted items. Next, all original nodes in TSG are softly clustered into factor nodes, each of which represents a certain factor of the user's preference. At last, all factor nodes' representations are used together to predict SR results. Our extensive experiments upon two datasets justify that our HGNN-based factorial user modeling obtains better SR performance than the state-of-the-art SR models.

Compared with traditional sentence-level relation extraction, document-level relation extraction is a more challenging task where an entity in a document may be mentioned multiple times and associated with multiple relations.However, most methods of documentlevel relation extraction do not distinguish between mention-level features and entity-level features, and just apply simple pooling operation for aggregating mention-level features into entity-level features.As a result, the distinct semantics between the different mentions of an entity are overlooked.To address this problem, we propose RSMAN in this paper which performs selective attentions over different entity mentions with respect to candidate relations.In this manner, the flexible and relation-specific representations of entities are obtained which indeed benefit relation classification.Our extensive experiments upon two benchmark datasets show that our RSMAN can bring significant improvements for some backbone models to achieve state-of-the-art performance, especially when an entity have multiple mentions in the document.1

Deep neural networks (DNNs) have been widely imported into collaborative-filtering (CF) based recommender systems and yielded remarkable superiority over traditional recommendation models. However, most deep CF-based models perform weakly when observed user-item interactions are sparse since user preferences and item characteristics are inferred mainly based on observed (historical) interactions. To address this problem, we propose a deep knowledge-enhanced recommendation model in this paper. Specifically, to augment user/item representations in the scenario of sparse historical user-item interactions, we first incorporate the knowledge from open knowledge graphs and personal information of users as side information, from which sufficient features of users and items are extracted. Second, to well capture shifted user preferences, we leverage a memory component constituted by recently interacted items rather than all historical ones. Third, attentive user representations are generated by attention mechanism to capture the diversity of user preferences. Furthermore, we build a convolutional neural network to pool the latent features in user representations for better user modeling, which enhances recommendation performance further. Our extensive experiments conducted against two real-world datasets, i.e., Douban movie and NetEase music, demonstrate our model’s remarkable superiority over the state-of-the-art deep recommendation models.

Li Cui, Deqing Yang, Jiaxin Yu, Chengwei Hu, Jiayang Cheng, Jingjie Yi, Yanghua Xiao. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Self-attention mechanism is primarily designed to capture the correlation (interaction) between any two objects in a sequence. Inspired by self-attention's success in many NLP tasks, some researchers have employed self-attention in sequential recommendation to refine user representations by capturing the correlations between the historical interacted items of a user. However, the user representations in previous self-attention based models are not flexible enough since the self-attention is only applied on user side, restricting performance improvement. In this paper, we propose a deep recommendation model with feature-level self-attention, namely SAFrec, which exhibits enhanced recommendation performance mainly due to its two advantages. The first one is that SAFrec employs self-attention mechanism on user side and item side simultaneously, to co-refine user representations and item representations. The second one is that, SAFrec leverages item features distilled from open knowledge graphs or websites, to represent users and items on fine-grained level (feature-level). Thus the correlations between users and items are discovered sufficiently. The extensive experiments conducted over two real datasets (NetEase music and Book-Crossing) not only demonstrate SAFrec's superiority on top-n recommendation over the state-of-the-art deep recommendation models, but also validate the significance of incorporating self-attention mechanism and feature-level representations.

Session-based recommendation (SR) has become an important and popular component of various e-commerce platforms, which aims to predict the next interacted item based on a given session. Most of existing SR models only focus on exploiting the consecutive items in a session interacted by a certain user, to capture the transition pattern among the items. Although some of them have been proven effective, the following two insights are often neglected. First, a user's micro-behaviors, such as the manner in which the user locates an item, the activities that the user commits on an item (e.g., reading comments, adding to cart), offer fine-grained and deep understanding of the user's preference. Second, the item attributes, also known as item knowledge, provide side information to model the transition pattern among interacted items and alleviate the data sparsity problem. These insights motivate us to propose a novel SR model MKM-SR in this paper, which incorporates user Micro-behaviors and item Knowledge into Multi-task learning for Session-based Recommendation. Specifically, a given session is modeled on micro-behavior level in MKM-SR, i.e., with a sequence of item-operation pairs rather than a sequence of items, to capture the transition pattern in the session sufficiently. Furthermore, we propose a multi-task learning paradigm to involve learning knowledge embeddings which plays a role as an auxiliary task to promote the major task of SR. It enables our model to obtain better session representations, resulting in more precise SR recommendation results. The extensive evaluations on two benchmark datasets demonstrate MKM-SR's superiority over the state-of-the-art SR models, justifying the strategy of incorporating knowledge learning.

Deep neural networks (DNNs) have been widely employed in recommender systems including incorporating attention mechanism for performance improvement. However, most of existing attention-based models only apply item-level attention on user side, restricting the further enhancement of recommendation performance. In this paper, we propose a knowledge-enhanced recommendation model ACAM, which incorporates item attributes distilled from knowledge graphs (KGs) as side information, and is built with a co-attention mechanism on attribute-level to achieve performance gains. Specifically, each user and item in ACAM are represented by a set of attribute embeddings at first. Then, user representations and item representations are augmented simultaneously through capturing the correlations between different attributes by a co-attention module. Our extensive experiments over two realistic datasets show that the user representations and item representations augmented by attribute-level co-attention gain ACAM's superiority over the state-of-the-art deep models.

Recently, many researchers in recommender systems have realized that encoding user-item interactions based on deep neural networks (DNNs) promotes collaborative-filtering (CF)'s performance. Nonetheless, those DNN-based models' performance is still limited when observed user-item interactions are very less because the training samples distilled from these interactions are critical for deep learning models. To address this problem, we resort to plenty features distilled from knowledge graphs (KGs), to profile users and items precisely and sufficiently rather than observed user-item interactions. In this paper, we propose a knowledge embedding based recommendation framework to alleviate the problem of sparse user-item interactions in recommendation. In our framework, each user and each item are both represented by the combination of an item embedding and a tag embedding at first. Specifically, item embeddings are learned by Metapath2Vec which is a graph embedding model qualified to embedding heterogeneous information networks. Tag embeddings are learned by a Skip-gram model similar to word embedding. We regarded these embeddings as knowledge embeddings because they both indicate knowledge about the latent relationships of movie-movie and user-movie. At last, a target user's representation and a candidate movie's representation are both fed into a multi-layer perceptron to output the probability that the user likes the item. The probability can be further used to achieve top-n recommendation. The extensive experiments on a movie recommendation dataset demonstrate our framework's superiority over some state-of-the-art recommendation models, especially in the scenario of sparse user-movie interactions.

Recommender systems have become indispensable for recommending items of interest to users and have been successfully deployed in a wide range of real-world applications. In this paper, we exploit the community influence of users in social networks to improve the recommendation accuracy. Depending on the community in which the user is located, the user's preferences are defined more accurately. Specifically, we propose a community-based collaborative filtering method for social recommender systems, which makes full use of the rich link/community structure within a social network. We first group users in a social network into overlapping communities. Then we explicitly incorporate the community preference into the latent factor model. Compared with seven state-of-the-art methods on four real-world datasets, our method achieves the best performance.

Explicitly exploring the semantics of a relation is significant for high-accuracy relation extraction, which is, however, not fully studied in previous work. In this paper, we mine the topic knowledge of a relation to explicitly represent the semantics of this relation, and model relation extraction as a matching problem. That is, the matching score between a sentence and a candidate relation is predicted for an entity pair. To this end, we propose a deep matching network to precisely model the semantic similarity between a sentence-relation pair. Besides, the topic knowledge also allows us to derive the importance information of samples as well as two knowledge-guided negative sampling strategies in the training process. We conduct extensive experiments to evaluate the proposed framework and observe improvements in AUC of 11.5% and max F1 of 5.4% over the baselines with state-of-the-art performance.

Most of existing embedding based recommendation models use embeddings (vectors) to represent users and items which contain latent features of users and items. Each of such embeddings corresponds to a single fixed point in low-dimensional space, thus fails to precisely represent the users/items with uncertainty which are often observed in recommender systems. Addressing this problem, we propose a unified deep recommendation framework employing Gaussian embeddings, which are proven adaptive to uncertain preferences exhibited by some users, resulting in better user representations and recommendation performance. Furthermore, our framework adopts Monte-Carlo sampling and convolutional neural networks to compute the correlation between the objective user and the candidate item, based on which precise recommendations are achieved. Our extensive experiments on two benchmark datasets not only justify that our proposed Gaussian embeddings capture the uncertainty of users very well, but also demonstrate its superior performance over the state-of-the-art recommendation models.

Most of traditional recommender systems perform well only when sufficient user-item interactions, such as purchase records or ratings, have been obtained in advance, while suffering from poor performance in the scenario of sparse interactions. Addressing this problem, we propose a neural network based recommendation framework which is fed with user/item'soriginal tags as well as the expanded tags from social context. Through embedding the latent correlations between tags into distributed feature representations, our model uncovers the implicit relationships between users and items sufficiently, exhibiting superior performance no matter whether sufficient user-item interactions are available or not. Furthermore, our framework can be further tailored for link prediction in networks, since recommending an item to a user can be recognized as predicting a link between them. The extensive experiments on two real recommendation tasks, i.e., Weibo followship recommendation and Douban movie recommendation, justify our framework's superiority to the state-of-the-art methods.

This paper addresses the problem ofmulti-instance entity typing from corpus. Current approaches mainly rely on the structured features (\textitattributes, attribute-value pairs andtags ) of the entities. However, their effectiveness is largely dependent on the completeness of structured features, which unfortunately is not guaranteed in KBs. In this paper, we therefore propose to use the text corpus of an entity to infer its types, and propose a multi-instance method to tackle this problem. We take each mention of an entity in KBs as an instance of the entity, and learn the types of these entities from multiple instances. Specifically, we first use an end-to-end neural network model to type each instance of an entity, and then use an integer linear programming (ILP) method to aggregate the predicted type results from multiple instances. Experimental results show the effectiveness of our method.

Although many researchers of recommender systems have noted that encoding user-item interactions based on DNNs promotes the performance of collaborative filtering, they ignore that embedding the latent features collected from external sources, e.g., knowledge graphs (KGs), is able to produce more precise recommendation results. Furthermore, CF-based models are still vulnerable to the scenarios of sparse known user-item interactions. In this paper, towards movie recommendation, we propose a novel knowledge-enhanced deep recommendation framework incorporating GAN-based models to acquire robust performance. Specifically, our framework first imports various feature embeddings distilled not only from user-movie interactions, but also from KGs and tags, to constitute initial user/movie representations. Then, user/movie representations are fed into a generator and a discriminator simultaneously to learn final optimal representations through adversarial training, which are conducive to generating better recommendation results. The extensive experiments on a real Douban dataset demonstrate our framework's superiority over some state-of-the-art recommendation models, especially in the scenarios of sparse observed user-movie interactions.

Frameworks are popularly used to reduce implementation complexity and improve productivity. Unfortunately, most frameworks are quite complex and not well documented. Hence, correctly and effectively programming with Framework is still a great challenge. One of the significant obstacles for us to smoothly use Framework is the complicated attribute value configuration of XML files. To overcome these difficulties, we present XMLValue to automatically recommend XML attribute values using association rules and NLP techniques. Experimental results show that our tool is efficient and effective for mining reusable configuration snippets, and has significantly shorten development time for framework based programming, and is general enough to support a variety of frameworks, and has real time performance for code assistance.

Almost all of the fish have the ability to recognize the underwater environment, among of which current forecasting is very critical for the movement of fish as well as an underwater vehicle. Perceiving water flow is inspired by the study on lateral line system of fish, and pressure sensor array is utilized to perceive changes in pressure of water. Due to the high dimension of sensor data, it is necessary to analyze the characteristics of the signal and obtain feature vectors to speed up the process of forecasting. Linear Discriminant Analysis (LDA) is proposed to preprocess the sensor data to obtain the compressed sampling of sensor data at first. Then, Support Vector Machine (SVM) is utilized to train a vigorous and robust classifier for recognizing the direction of flow. The forecast is evaluated according to the effectiveness and the prediction performance of the established pattern recognition model, which is able to recognize flow direction accurately and estimate the magnitude within a certain range like animals. Results indicate that the processed method not only can forecast flow direction, but also can combine with IMU sensor to estimate the attitude of the vehicle.

Who are the best targets to receive a call-for-paper or call-for-participation? What kind of topics should we propose for a workshop or a special issue of next year? Precisely predicting author's topic following behavior, i.e., publishing papers of a certain research topic in future, is essential to answer these questions. In this paper, we aim to model and predict author's topic following behavior in a heterogeneous information network. The heart of our methodology is to evaluate the author-author similarity through informative meta paths in the network. The models we propose in this paper can predict not only whether a given author will follow a certain topic but also the topic distribution over all publications in the next year. Extensive experimental evaluations justify that the prediction performance of our approach outperforms the existing approaches across various topics.

Given the users from a social network site, who have been tagged with a set of terms, how can we recommend the movies tagged with a completely different set of terms hosted by another website? Given the users from a website dedicated to Type I and Type II diabetes, how can we recommend the discussion threads from another website dedicated to gestational diabetes, where the keywords used in the two websites might be quite diverse? In other words, how can we recommend across heterogeneous domains characterized by barely overlapping feature sets?

Cross-domain recommendation has attracted wide research interest which generally aims at improving the recommendation performance by alleviating the cold start problem in collaborative filtering based recommendation or generating a more comprehensive user profiles from multiple domains. In most previous cross-domain recommendation settings, explicit or implicit relationships can be easily established across different domains. However, many real applications belong to a more challenging setting: recommendation across heterogeneous domains without explicit relationships, where neither explicit user-item relations nor overlapping features exist between different domains. In this new setting, we need to (1) enrich the sparse data to characterize users or items and (2) bridge the gap caused by the heterogenous features in different domains. To overcome the first challenge, we proposed an optimized local tag propagation algorithm to generate descriptive tags for user profiling. For the second challenge, we proposed a semantic relatedness metric by mapping the heterogenous features onto their concept space derived from online encyclopedias. We conducted extensive experiments on two real datasets to justify the effectiveness of our solution.