Zhixu Li

Qianyu He, Sijie Cheng, Zhixu Li et al. · tsinghua

Simile interpretation is a crucial task in natural language processing. Nowadays, pre-trained language models (PLMs) have achieved state-of-the-art performance on many tasks. However, it remains under-explored whether PLMs can interpret similes or not. In this paper, we investigate the ability of PLMs in simile interpretation by designing a novel task named Simile Property Probing, i.e., to let the PLMs infer the shared properties of similes. We construct our simile property probing datasets from both general textual corpora and human-designed questions, containing 1,633 examples covering seven main categories. Our empirical study based on the constructed datasets shows that PLMs can infer similes' shared properties while still underperforming humans. To bridge the gap with human performance, we additionally design a knowledge-enhanced training objective by incorporating the simile knowledge into PLMs via knowledge embedding methods. Our method results in a gain of 8.58% in the probing task and 1.37% in the downstream task of sentiment classification. The datasets and code are publicly available at https://github.com/Abbey4799/PLMs-Interpret-Simile.

Jiaan Wang, Yunlong Liang, Fandong Meng et al. · tsinghua

Recently, the emergence of ChatGPT has attracted wide attention from the computational linguistics community. Many prior studies have shown that ChatGPT achieves remarkable performance on various NLP tasks in terms of automatic evaluation metrics. However, the ability of ChatGPT to serve as an evaluation metric is still underexplored. Considering assessing the quality of natural language generation (NLG) models is an arduous task and NLG metrics notoriously show their poor correlation with human judgments, we wonder whether ChatGPT is a good NLG evaluation metric. In this report, we provide a preliminary meta-evaluation on ChatGPT to show its reliability as an NLG metric. In detail, we regard ChatGPT as a human evaluator and give task-specific (e.g., summarization) and aspect-specific (e.g., relevance) instruction to prompt ChatGPT to evaluate the generated results of NLG models. We conduct experiments on five NLG meta-evaluation datasets (including summarization, story generation and data-to-text tasks). Experimental results show that compared with previous automatic metrics, ChatGPT achieves state-of-the-art or competitive correlation with human judgments in most cases. In addition, we find that the effectiveness of the ChatGPT evaluator might be influenced by the creation method of the meta-evaluation datasets. For the meta-evaluation datasets which are created greatly depending on the reference and thus are biased, the ChatGPT evaluator might lose its effectiveness. We hope our preliminary study could prompt the emergence of a general-purposed reliable NLG metric.

Xuwu Wang, Junfeng Tian, Min Gui et al.

Multimodal Entity Linking (MEL) which aims at linking mentions with multimodal contexts to the referent entities from a knowledge base (e.g., Wikipedia), is an essential task for many multimodal applications. Although much attention has been paid to MEL, the shortcomings of existing MEL datasets including limited contextual topics and entity types, simplified mention ambiguity, and restricted availability, have caused great obstacles to the research and application of MEL. In this paper, we present WikiDiverse, a high-quality human-annotated MEL dataset with diversified contextual topics and entity types from Wikinews, which uses Wikipedia as the corresponding knowledge base. A well-tailored annotation procedure is adopted to ensure the quality of the dataset. Based on WikiDiverse, a sequence of well-designed MEL models with intra-modality and inter-modality attentions are implemented, which utilize the visual information of images more adequately than existing MEL models do. Extensive experimental analyses are conducted to investigate the contributions of different modalities in terms of MEL, facilitating the future research on this task. The dataset and baseline models are available at https://github.com/wangxw5/wikiDiverse.

Jiaan Wang, Fandong Meng, Duo Zheng et al. · tsinghua

Cross-lingual summarization is the task of generating a summary in one language (e.g., English) for the given document(s) in a different language (e.g., Chinese). Under the globalization background, this task has attracted increasing attention of the computational linguistics community. Nevertheless, there still remains a lack of comprehensive review for this task. Therefore, we present the first systematic critical review on the datasets, approaches, and challenges in this field. Specifically, we carefully organize existing datasets and approaches according to different construction methods and solution paradigms, respectively. For each type of datasets or approaches, we thoroughly introduce and summarize previous efforts and further compare them with each other to provide deeper analyses. In the end, we also discuss promising directions and offer our thoughts to facilitate future research. This survey is for both beginners and experts in cross-lingual summarization, and we hope it will serve as a starting point as well as a source of new ideas for researchers and engineers interested in this area.

Jiaan Wang, Yunlong Liang, Fandong Meng et al. · tsinghua

Given a document in a source language, cross-lingual summarization (CLS) aims to generate a summary in a different target language. Recently, the emergence of Large Language Models (LLMs), such as GPT-3.5, ChatGPT and GPT-4, has attracted wide attention from the computational linguistics community. However, it is not yet known the performance of LLMs on CLS. In this report, we empirically use various prompts to guide LLMs to perform zero-shot CLS from different paradigms (i.e., end-to-end and pipeline), and provide a preliminary evaluation on the generated summaries. We find that ChatGPT and GPT-4 originally prefer to produce lengthy summaries with detailed information. These two LLMs can further balance informativeness and conciseness with the help of an interactive prompt, significantly improving their CLS performance. Experimental results on three widely-used CLS datasets show that GPT-4 achieves state-of-the-art zero-shot CLS performance, and performs competitively compared with the fine-tuned mBART-50. Moreover, we also find some multi-lingual and bilingual LLMs (i.e., BLOOMZ, ChatGLM-6B, Vicuna-13B and ChatYuan) have limited zero-shot CLS ability. Due to the composite nature of CLS, which requires models to perform summarization and translation simultaneously, accomplishing this task in a zero-shot manner is even a challenge for LLMs. Therefore, we sincerely hope and recommend future LLM research could use CLS as a testbed.

Siyu Yuan, Deqing Yang, Jiaqing Liang et al.

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 fine-tuning upon the entity typing dataset. Besides, we only have heterogeneous training data consisting of a small portion of human-annotated data and a large portion of auto-generated but low-quality data. To tackle these problems, we employ curriculum learning (CL) to train our GET model upon the 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-of-the-art entity typing models. The code has been released on https://github.com/siyuyuan/GET.

Jiaan Wang, Fandong Meng, Yunlong Liang et al. · tsinghua

Given a document in a source language, cross-lingual summarization (CLS) aims at generating a concise summary in a different target language. Unlike monolingual summarization (MS), naturally occurring source-language documents paired with target-language summaries are rare. To collect large-scale CLS data, existing datasets typically involve translation in their creation. However, the translated text is distinguished from the text originally written in that language, i.e., translationese. In this paper, we first confirm that different approaches of constructing CLS datasets will lead to different degrees of translationese. Then we systematically investigate how translationese affects CLS model evaluation and performance when it appears in source documents or target summaries. In detail, we find that (1) the translationese in documents or summaries of test sets might lead to the discrepancy between human judgment and automatic evaluation; (2) the translationese in training sets would harm model performance in real-world applications; (3) though machine-translated documents involve translationese, they are very useful for building CLS systems on low-resource languages under specific training strategies. Lastly, we give suggestions for future CLS research including dataset and model developments. We hope that our work could let researchers notice the phenomenon of translationese in CLS and take it into account in the future.

Yiyou Sun, Xinyang Han, Weichen Zhang et al.

Recent AI systems have achieved strong results on a wide range of benchmarks, yet these gains have not translated into economically meaningful deployment across many professional domains. We argue that this gap is largely an evaluation problem: widely used benchmarks lack sustained performance measurement on real and economically valuable workflows. This paper introduces Agents' Last Exam (ALE), a benchmark designed to evaluate AI agents on long-horizon, economically valuable, real-world tasks with verifiable outcomes. Developed in collaboration with 250+ industry experts, ALE covers non-physical industries defined with reference to O*NET / SOC 2018 (the U.S. federal occupational taxonomy). It is organized around a task taxonomy with 55 subfields grouped into 13 industry clusters covering 1K+ tasks. Current results show that the hardest tier remains far from saturated: across mainstream harness and backbone configurations, the average full pass rate is 2.6%. ALE is designed as a living benchmark: its task pool grows continuously as new workflows and industries are onboarded. More broadly, ALE is intended not merely as another leaderboard, but as an instrument for closing the gap between benchmark success and GDP-relevant impact.

Yuyan Chen, Qiang Fu, Ge Fan et al.

Recent years, Pre-trained Language models (PLMs) have swept into various fields of artificial intelligence and achieved great success. However, most PLMs, such as T5 and GPT3, have a huge amount of parameters, fine-tuning them is often expensive and time consuming, and storing them takes up a lot of space. Therefore, it is necessary to adopt a parameter-efficient approach to reduce parameters of PLMs in fine-tuning without compromising their performance in downstream tasks. In this paper, we design a novel adapter which only acts on self-attention outputs in PLMs. This adapter adopts element-wise linear transformation using Hadamard product, hence named as Hadamard adapter, requires the fewest parameters compared to previous parameter-efficient adapters. In addition, we also summarize some tuning patterns for Hadamard adapter shared by various downstream tasks, expecting to provide some guidance for further parameter reduction with shared adapters in future studies. The experiments conducted on the widely-used GLUE benchmark with several SOTA PLMs prove that the Hadamard adapter achieves competitive performance with only 0.033\% parameters compared with full fine-tuning, and it has the fewest parameters compared with other adapters. Moreover, we further find that there is also some redundant layers in the Hadamard adapter which can be removed to achieve more parameter efficiency with only 0.022\% parameters.

Yaoxian Song, Penglei Sun, Piaopiao Jin et al.

Robotic grasping is a fundamental ability for a robot to interact with the environment. Current methods focus on how to obtain a stable and reliable grasping pose in object level, while little work has been studied on part (shape)-wise grasping which is related to fine-grained grasping and robotic affordance. Parts can be seen as atomic elements to compose an object, which contains rich semantic knowledge and a strong correlation with affordance. However, lacking a large part-wise 3D robotic dataset limits the development of part representation learning and downstream applications. In this paper, we propose a new large Language-guided SHape grAsPing datasEt (named LangSHAPE) to promote 3D part-level affordance and grasping ability learning. From the perspective of robotic cognition, we design a two-stage fine-grained robotic grasping framework (named LangPartGPD), including a novel 3D part language grounding model and a part-aware grasp pose detection model, in which explicit language input from human or large language models (LLMs) could guide a robot to generate part-level 6-DoF grasping pose with textual explanation. Our method combines the advantages of human-robot collaboration and LLMs' planning ability using explicit language as a symbolic intermediate. To evaluate the effectiveness of our proposed method, we perform 3D part grounding and fine-grained grasp detection experiments on both simulation and physical robot settings, following language instructions across different degrees of textual complexity. Results show our method achieves competitive performance in 3D geometry fine-grained grounding, object affordance inference, and 3D part-aware grasping tasks. Our dataset and code are available on our project website https://sites.google.com/view/lang-shape

Yuyan Chen, Zhixu Li, Jiaqing Liang et al.

Humor understanding is an important and challenging research in natural language processing. As the popularity of pre-trained language models (PLMs), some recent work makes preliminary attempts to adopt PLMs for humor recognition and generation. However, these simple attempts do not substantially answer the question: {\em whether PLMs are capable of humor understanding?} This paper is the first work that systematically investigates the humor understanding ability of PLMs. For this purpose, a comprehensive framework with three evaluation steps and four evaluation tasks is designed. We also construct a comprehensive Chinese humor dataset, which can fully meet all the data requirements of the proposed evaluation framework. Our empirical study on the Chinese humor dataset yields some valuable observations, which are of great guiding value for future optimization of PLMs in humor understanding and generation.

Jingdan Zhang, Jiaan Wang, Xiaodan Wang et al.

Multi-modal knowledge graphs (MMKGs) combine different modal data (e.g., text and image) for a comprehensive understanding of entities. Despite the recent progress of large-scale MMKGs, existing MMKGs neglect the multi-aspect nature of entities, limiting the ability to comprehend entities from various perspectives. In this paper, we construct AspectMMKG, the first MMKG with aspect-related images by matching images to different entity aspects. Specifically, we collect aspect-related images from a knowledge base, and further extract aspect-related sentences from the knowledge base as queries to retrieve a large number of aspect-related images via an online image search engine. Finally, AspectMMKG contains 2,380 entities, 18,139 entity aspects, and 645,383 aspect-related images. We demonstrate the usability of AspectMMKG in entity aspect linking (EAL) downstream task and show that previous EAL models achieve a new state-of-the-art performance with the help of AspectMMKG. To facilitate the research on aspect-related MMKG, we further propose an aspect-related image retrieval (AIR) model, that aims to correct and expand aspect-related images in AspectMMKG. We train an AIR model to learn the relationship between entity image and entity aspect-related images by incorporating entity image, aspect, and aspect image information. Experimental results indicate that the AIR model could retrieve suitable images for a given entity w.r.t different aspects.

Shaohui Zheng, Zhixu Li, Jiaan Wang et al.

Cross-Lingual Summarization (CLS) aims at generating summaries in one language for the given documents in another language. CLS has attracted wide research attention due to its practical significance in the multi-lingual world. Though great contributions have been made, existing CLS works typically focus on short documents, such as news articles, short dialogues and guides. Different from these short texts, long documents such as academic articles and business reports usually discuss complicated subjects and consist of thousands of words, making them non-trivial to process and summarize. To promote CLS research on long documents, we construct Perseus, the first long-document CLS dataset which collects about 94K Chinese scientific documents paired with English summaries. The average length of documents in Perseus is more than two thousand tokens. As a preliminary study on long-document CLS, we build and evaluate various CLS baselines, including pipeline and end-to-end methods. Experimental results on Perseus show the superiority of the end-to-end baseline, outperforming the strong pipeline models equipped with sophisticated machine translation systems. Furthermore, to provide a deeper understanding, we manually analyze the model outputs and discuss specific challenges faced by current approaches. We hope that our work could benchmark long-document CLS and benefit future studies.

Yuyan Chen, Qiang Fu, Yichen Yuan et al.

Large Language Models (LLMs) have gained widespread adoption in various natural language processing tasks, including question answering and dialogue systems. However, a major drawback of LLMs is the issue of hallucination, where they generate unfaithful or inconsistent content that deviates from the input source, leading to severe consequences. In this paper, we propose a robust discriminator named RelD to effectively detect hallucination in LLMs' generated answers. RelD is trained on the constructed RelQA, a bilingual question-answering dialogue dataset along with answers generated by LLMs and a comprehensive set of metrics. Our experimental results demonstrate that the proposed RelD successfully detects hallucination in the answers generated by diverse LLMs. Moreover, it performs well in distinguishing hallucination in LLMs' generated answers from both in-distribution and out-of-distribution datasets. Additionally, we also conduct a thorough analysis of the types of hallucinations that occur and present valuable insights. This research significantly contributes to the detection of reliable answers generated by LLMs and holds noteworthy implications for mitigating hallucination in the future work.

Kexin Wang, Zhixu Li, Jiaan Wang et al.

Grounding dialogue system with external knowledge is a promising way to improve the quality of responses. Most existing works adopt knowledge graphs (KGs) as the external resources, paying attention to the contribution of entities in the last utterance of the dialogue for context understanding and response generation. Nevertheless, the correlations between knowledge implied in the multi-turn context and the transition regularities between relations in KGs are under-explored. To this end, we propose a Relation Transition aware Knowledge-Grounded Dialogue Generation model (RT-KGD). Specifically, inspired by the latent logic of human conversation, our model integrates dialogue-level relation transition regularities with turn-level entity semantic information. In this manner, the interaction between knowledge is considered to produce abundant clues for predicting the appropriate knowledge and generating coherent responses. The experimental results on both automatic evaluation and manual evaluation indicate that our model outperforms state-of-the-art baselines.

Jiaan Wang, Jianfeng Qu, Yunlong Liang et al.

Constructing commonsense knowledge graphs (CKGs) has attracted wide research attention due to its significant importance in cognitive intelligence. Nevertheless, existing CKGs are typically oriented to English, limiting the research in non-English languages. Meanwhile, the emergence of foundation models like ChatGPT and GPT-4 has shown promising intelligence with the help of reinforcement learning from human feedback. Under the background, in this paper, we utilize foundation models to construct a Chinese CKG, named Snowman. Specifically, we distill different types of commonsense head items from ChatGPT, and continue to use it to collect tail items with respect to the head items and pre-defined relations. Based on the preliminary analysis, we find the negative commonsense knowledge distilled by ChatGPT achieves lower human acceptance compared to other knowledge. Therefore, we design a simple yet effective self-instruct filtering strategy to filter out invalid negative commonsense. Overall, the constructed Snowman covers more than ten million Chinese commonsense triples, making it the largest Chinese CKG. Moreover, human studies show the acceptance of Snowman achieves 90.6\%, indicating the high-quality triples distilled by the cutting-edge foundation model. We also conduct experiments on commonsense knowledge models to show the usability and effectiveness of our Snowman.

Wenhao Huang, Jiaqing Liang, Zhixu Li et al.

Information extraction (IE) has been studied extensively. The existing methods always follow a fixed extraction order for complex IE tasks with multiple elements to be extracted in one instance such as event extraction. However, we conduct experiments on several complex IE datasets and observe that different extraction orders can significantly affect the extraction results for a great portion of instances, and the ratio of sentences that are sensitive to extraction orders increases dramatically with the complexity of the IE task. Therefore, this paper proposes a novel adaptive ordered IE paradigm to find the optimal element extraction order for different instances, so as to achieve the best extraction results. We also propose an reinforcement learning (RL) based framework to generate optimal extraction order for each instance dynamically. Additionally, we propose a co-training framework adapted to RL to mitigate the exposure bias during the extractor training phase. Extensive experiments conducted on several public datasets demonstrate that our proposed method can beat previous methods and effectively improve the performance of various IE tasks, especially for complex ones.

Zhouhong Gu, Sihang Jiang, Jingping Liu et al.

Taxonomy is formulated as directed acyclic concepts graphs or trees that support many downstream tasks. Many new coming concepts need to be added to an existing taxonomy. The traditional taxonomy expansion task aims only at finding the best position for new coming concepts in the existing taxonomy. However, they have two drawbacks when being applied to the real-scenarios. The previous methods suffer from low-efficiency since they waste much time when most of the new coming concepts are indeed noisy concepts. They also suffer from low-effectiveness since they collect training samples only from the existing taxonomy, which limits the ability of the model to mine more hypernym-hyponym relationships among real concepts. This paper proposes a pluggable framework called Generative Adversarial Network for Taxonomy Entering Evaluation (GANTEE) to alleviate these drawbacks. A generative adversarial network is designed in this framework by discriminative models to alleviate the first drawback and the generative model to alleviate the second drawback. Two discriminators are used in GANTEE to provide long-term and short-term rewards, respectively. Moreover, to further improve the efficiency, pre-trained language models are used to retrieve the representation of the concepts quickly. The experiments on three real-world large-scale datasets with two different languages show that GANTEE improves the performance of the existing taxonomy expansion methods in both effectiveness and efficiency.

Yuyan Chen, Zhihao Wen, Ge Fan et al.

Prompt engineering, as an efficient and effective way to leverage Large Language Models (LLM), has drawn a lot of attention from the research community. The existing research primarily emphasizes the importance of adapting prompts to specific tasks, rather than specific LLMs. However, a good prompt is not solely defined by its wording, but also binds to the nature of the LLM in question. In this work, we first quantitatively demonstrate that different prompts should be adapted to different LLMs to enhance their capabilities across various downstream tasks in NLP. Then we novelly propose a model-adaptive prompt optimizer (MAPO) method that optimizes the original prompts for each specific LLM in downstream tasks. Extensive experiments indicate that the proposed method can effectively refine prompts for an LLM, leading to significant improvements over various downstream tasks.

Yuyan Chen, Songzhou Yan, Zhihong Zhu et al.

Humor, deeply rooted in societal meanings and cultural details, poses a unique challenge for machines. While advances have been made in natural language processing, real-world humor often thrives in a multi-modal context, encapsulated distinctively by memes. This paper poses a particular emphasis on the impact of multi-images on meme captioning. After that, we introduce the \textsc{XMeCap} framework, a novel approach that adopts supervised fine-tuning and reinforcement learning based on an innovative reward model, which factors in both global and local similarities between visuals and text. Our results, benchmarked against contemporary models, manifest a marked improvement in caption generation for both single-image and multi-image memes, as well as different meme categories. \textsc{XMeCap} achieves an average evaluation score of 75.85 for single-image memes and 66.32 for multi-image memes, outperforming the best baseline by 6.75\% and 8.56\%, respectively. This research not only establishes a new frontier in meme-related studies but also underscores the potential of machines in understanding and generating humor in a multi-modal setting.

Yuyan Chen, Yiwen Qian, Songzhou Yan et al.

In the era of social media video platforms, popular ``hot-comments'' play a crucial role in attracting user impressions of short-form videos, making them vital for marketing and branding purpose. However, existing research predominantly focuses on generating descriptive comments or ``danmaku'' in English, offering immediate reactions to specific video moments. Addressing this gap, our study introduces \textsc{HotVCom}, the largest Chinese video hot-comment dataset, comprising 94k diverse videos and 137 million comments. We also present the \texttt{ComHeat} framework, which synergistically integrates visual, auditory, and textual data to generate influential hot-comments on the Chinese video dataset. Empirical evaluations highlight the effectiveness of our framework, demonstrating its excellence on both the newly constructed and existing datasets.

Penglei Sun, Yaoxian Song, Xinglin Pan et al.

The existing works on object-level language grounding with 3D objects mostly focus on improving performance by utilizing the off-the-shelf pre-trained models to capture features, such as viewpoint selection or geometric priors. However, they have failed to consider exploring the cross-modal representation of language-vision alignment in the cross-domain field. To answer this problem, we propose a novel method called Domain Adaptation for Language Grounding (DA4LG) with 3D objects. Specifically, the proposed DA4LG consists of a visual adapter module with multi-task learning to realize vision-language alignment by comprehensive multimodal feature representation. Experimental results demonstrate that DA4LG competitively performs across visual and non-visual language descriptions, independent of the completeness of observation. DA4LG achieves state-of-the-art performance in the single-view setting and multi-view setting with the accuracy of 83.8% and 86.8% respectively in the language grounding benchmark SNARE. The simulation experiments show the well-practical and generalized performance of DA4LG compared to the existing methods. Our project is available at https://sites.google.com/view/da4lg.

Wenhao Huang, Zhouhong Gu, Chenghao Peng et al.

Web scraping is a powerful technique that extracts data from websites, enabling automated data collection, enhancing data analysis capabilities, and minimizing manual data entry efforts. Existing methods, wrappers-based methods suffer from limited adaptability and scalability when faced with a new website, while language agents, empowered by large language models (LLMs), exhibit poor reusability in diverse web environments. In this work, we introduce the paradigm of generating web scrapers with LLMs and propose AutoScraper, a two-stage framework that can handle diverse and changing web environments more efficiently. AutoScraper leverages the hierarchical structure of HTML and similarity across different web pages for generating web scrapers. Besides, we propose a new executability metric for better measuring the performance of web scraper generation tasks. We conduct comprehensive experiments with multiple LLMs and demonstrate the effectiveness of our framework. Resources of this paper can be found at \url{https://github.com/EZ-hwh/AutoScraper}

Xiangru Zhu, Penglei Sun, Chengyu Wang et al.

Text-to-image (T2I) synthesis has recently achieved significant advancements. However, challenges remain in the model's compositionality, which is the ability to create new combinations from known components. We introduce Winoground-T2I, a benchmark designed to evaluate the compositionality of T2I models. This benchmark includes 11K complex, high-quality contrastive sentence pairs spanning 20 categories. These contrastive sentence pairs with subtle differences enable fine-grained evaluations of T2I synthesis models. Additionally, to address the inconsistency across different metrics, we propose a strategy that evaluates the reliability of various metrics by using comparative sentence pairs. We use Winoground-T2I with a dual objective: to evaluate the performance of T2I models and the metrics used for their evaluation. Finally, we provide insights into the strengths and weaknesses of these metrics and the capabilities of current T2I models in tackling challenges across a range of complex compositional categories. Our benchmark is publicly available at https://github.com/zhuxiangru/Winoground-T2I .

Xiangru Zhu, Penglei Sun, Yaoxian Song et al.

Accurate interpretation and visualization of human instructions are crucial for text-to-image (T2I) synthesis. However, current models struggle to capture semantic variations from word order changes, and existing evaluations, relying on indirect metrics like text-image similarity, fail to reliably assess these challenges. This often obscures poor performance on complex or uncommon linguistic patterns by the focus on frequent word combinations. To address these deficiencies, we propose a novel metric called SemVarEffect and a benchmark named SemVarBench, designed to evaluate the causality between semantic variations in inputs and outputs in T2I synthesis. Semantic variations are achieved through two types of linguistic permutations, while avoiding easily predictable literal variations. Experiments reveal that the CogView-3-Plus and Ideogram 2 performed the best, achieving a score of 0.2/1. Semantic variations in object relations are less understood than attributes, scoring 0.07/1 compared to 0.17-0.19/1. We found that cross-modal alignment in UNet or Transformers plays a crucial role in handling semantic variations, a factor previously overlooked by a focus on textual encoders. Our work establishes an effective evaluation framework that advances the T2I synthesis community's exploration of human instruction understanding. Our benchmark and code are available at https://github.com/zhuxiangru/SemVarBench .

Wenhao Huang, Qianyu He, Zhixu Li et al.

Definition bias is a negative phenomenon that can mislead models. Definition bias in information extraction appears not only across datasets from different domains but also within datasets sharing the same domain. We identify two types of definition bias in IE: bias among information extraction datasets and bias between information extraction datasets and instruction tuning datasets. To systematically investigate definition bias, we conduct three probing experiments to quantitatively analyze it and discover the limitations of unified information extraction and large language models in solving definition bias. To mitigate definition bias in information extraction, we propose a multi-stage framework consisting of definition bias measurement, bias-aware fine-tuning, and task-specific bias mitigation. Experimental results demonstrate the effectiveness of our framework in addressing definition bias. Resources of this paper can be found at https://github.com/EZ-hwh/definition-bias

Haiquan Zhao, Lingyu Li, Shisong Chen et al.

Emotion Support Conversation (ESC) is a crucial application, which aims to reduce human stress, offer emotional guidance, and ultimately enhance human mental and physical well-being. With the advancement of Large Language Models (LLMs), many researchers have employed LLMs as the ESC models. However, the evaluation of these LLM-based ESCs remains uncertain. Inspired by the awesome development of role-playing agents, we propose an ESC Evaluation framework (ESC-Eval), which uses a role-playing agent to interact with ESC models, followed by a manual evaluation of the interactive dialogues. In detail, we first re-organize 2,801 role-playing cards from seven existing datasets to define the roles of the role-playing agent. Second, we train a specific role-playing model called ESC-Role which behaves more like a confused person than GPT-4. Third, through ESC-Role and organized role cards, we systematically conduct experiments using 14 LLMs as the ESC models, including general AI-assistant LLMs (ChatGPT) and ESC-oriented LLMs (ExTES-Llama). We conduct comprehensive human annotations on interactive multi-turn dialogues of different ESC models. The results show that ESC-oriented LLMs exhibit superior ESC abilities compared to general AI-assistant LLMs, but there is still a gap behind human performance. Moreover, to automate the scoring process for future ESC models, we developed ESC-RANK, which trained on the annotated data, achieving a scoring performance surpassing 35 points of GPT-4. Our data and code are available at https://github.com/AIFlames/Esc-Eval.

Jiaan Wang, Fandong Meng, Ziyao Lu et al.

We present ClidSum, a benchmark dataset for building cross-lingual summarization systems on dialogue documents. It consists of 67k+ dialogue documents from two subsets (i.e., SAMSum and MediaSum) and 112k+ annotated summaries in different target languages. Based on the proposed ClidSum, we introduce two benchmark settings for supervised and semi-supervised scenarios, respectively. We then build various baseline systems in different paradigms (pipeline and end-to-end) and conduct extensive experiments on ClidSum to provide deeper analyses. Furthermore, we propose mDialBART which extends mBART-50 (a multi-lingual BART) via further pre-training. The multiple objectives used in the further pre-training stage help the pre-trained model capture the structural characteristics as well as important content in dialogues and the transformation from source to the target language. Experimental results show the superiority of mDialBART, as an end-to-end model, outperforms strong pipeline models on ClidSum. Finally, we discuss specific challenges that current approaches faced with this task and give multiple promising directions for future research. We have released the dataset and code at https://github.com/krystalan/ClidSum.

Xiaopan Zhang, Zejin Wang, Zhixu Li et al.

To complete assignments provided by humans in natural language, robots must interpret commands, generate and answer relevant questions for scene understanding, and manipulate target objects. Real-world deployments often require multiple heterogeneous robots with different manipulation capabilities to handle different assignments cooperatively. Beyond the need for specialized manipulation skills, effective information gathering is important in completing these assignments. To address this component of the problem, we formalize the information-gathering process in a fully cooperative setting as an underexplored multi-agent multi-task Embodied Question Answering (MM-EQA) problem, which is a novel extension of canonical Embodied Question Answering (EQA), where effective communication is crucial for coordinating efforts without redundancy. To address this problem, we propose CommCP, a novel LLM-based decentralized communication framework designed for MM-EQA. Our framework employs conformal prediction to calibrate the generated messages, thereby minimizing receiver distractions and enhancing communication reliability. To evaluate our framework, we introduce an MM-EQA benchmark featuring diverse, photo-realistic household scenarios with embodied questions. Experimental results demonstrate that CommCP significantly enhances the task success rate and exploration efficiency over baselines. The experiment videos, code, and dataset are available on our project website: https://comm-cp.github.io.

Chunyan An, Yunhan Li, Qiang Yang et al.

Session-based Social Recommendation (SSR) leverages social relationships within online networks to enhance the performance of Session-based Recommendation (SR). However, existing SSR algorithms often encounter the challenge of "friend data sparsity". Moreover, significant discrepancies can exist between the purchase preferences of social network friends and those of the target user, reducing the influence of friends relative to the target user's own preferences. To address these challenges, this paper introduces the concept of "Like-minded Peers" (LMP), representing users whose preferences align with the target user's current session based on their historical sessions. This is the first work, to our knowledge, that uses LMP to enhance the modeling of social influence in SSR. This approach not only alleviates the problem of friend data sparsity but also effectively incorporates users with similar preferences to the target user. We propose a novel model named Transformer Encoder with Graph Attention Aggregator Recommendation (TEGAARec), which includes the TEGAA module and the GAT-based social aggregation module. The TEGAA module captures and merges both long-term and short-term interests for target users and LMP users. Concurrently, the GAT-based social aggregation module is designed to aggregate the target users' dynamic interests and social influence in a weighted manner. Extensive experiments on four real-world datasets demonstrate the efficacy and superiority of our proposed model and ablation studies are done to illustrate the contributions of each component in TEGAARec.

Jiaan Wang, Jianfeng Qu, Kexin Wang et al.

Knowledge-grounded dialogue (KGD) learns to generate an informative response based on a given dialogue context and external knowledge (\emph{e.g.}, knowledge graphs; KGs). Recently, the emergence of large language models (LLMs) and pre-training techniques has brought great success to knowledge-grounded dialogue. However, when building KGD systems in real applications, there are various real-world noises that are inevitable to face. For example, the dialogue context might involve perturbations such as misspellings and abbreviations. In addition, KGs typically suffer from incompletion and also might contain erroneous and outdated facts. Such real-world noises pose a challenge to the robustness of KGD systems and hinder their applications in the real world. In this paper, we propose an entity-based contrastive learning framework for improving the robustness of KGD. Specifically, we make use of the entity information in a KGD sample to create both its positive and negative samples which involve semantic-irrelevant and semantic-relevant perturbations, respectively. The contrastive learning framework ensures the KGD model is aware of these two types of perturbations, thus generating informative responses with the potentially noisy inputs in real applications. Experimental results on three benchmark datasets show that our method achieves new state-of-the-art performance in terms of automatic evaluation scores, verifying its effectiveness and potentiality. Furthermore, we show that our method can generate better responses than comparison models in both the noisy and the few-shot settings.

Chenghua Huang, Lu Wang, Fangkai Yang et al.

Proximal Policy Optimization (PPO)-based Reinforcement Learning from Human Feedback (RLHF) is essential for aligning large language models (LLMs) with human preferences. It requires joint training of an actor and critic with a pretrained, fixed reward model for guidance. This approach increases computational complexity and instability due to actor-critic interdependence. Additionally, PPO lacks access to true environment rewards in LLM tasks, limiting its adaptability. Under such conditions, pretraining a value model or a reward model becomes equivalent, as both provide fixed supervisory signals without new ground-truth feedback. To address these issues, we propose \textbf{Decoupled Value Policy Optimization (DVPO)}, a lean framework that replaces traditional reward modeling with a pretrained \emph{global value model (GVM)}. The GVM is conditioned on policy trajectories and predicts token-level return-to-go estimates. By decoupling value model from policy training (via frozen GVM-driven RL objectives), DVPO eliminates actor-critic interdependence, reducing GPU memory usage by 40\% and training time by 35\% compared to conventional RLHF. Experiments across benchmarks show DVPO outperforms efficient RLHF methods (e.g., DPO) while matching state-of-the-art PPO in performance.

Haiquan Zhao, Xuwu Wang, Shisong Chen et al.

In recent years, multi-modal entity linking (MEL) has garnered increasing attention in the research community due to its significance in numerous multi-modal applications. Video, as a popular means of information transmission, has become prevalent in people's daily lives. However, most existing MEL methods primarily focus on linking textual and visual mentions or offline videos's mentions to entities in multi-modal knowledge bases, with limited efforts devoted to linking mentions within online video content. In this paper, we propose a task called Online Video Entity Linking OVEL, aiming to establish connections between mentions in online videos and a knowledge base with high accuracy and timeliness. To facilitate the research works of OVEL, we specifically concentrate on live delivery scenarios and construct a live delivery entity linking dataset called LIVE. Besides, we propose an evaluation metric that considers timelessness, robustness, and accuracy. Furthermore, to effectively handle OVEL task, we leverage a memory block managed by a Large Language Model and retrieve entity candidates from the knowledge base to augment LLM performance on memory management. The experimental results prove the effectiveness and efficiency of our method.

Zhiwei Zha, Jiaan Wang, Zhixu Li et al.

Multimodal knowledge bases (MMKBs) provide cross-modal aligned knowledge crucial for multimodal tasks. However, the images in existing MMKBs are generally collected for entities in encyclopedia knowledge graphs. Therefore, detailed groundings of visual semantics with linguistic concepts are lacking, which are essential for the visual concept cognition ability of multimodal models. Addressing this gap, we introduce M^2ConceptBase, the first concept-centric MMKB. M^2ConceptBase models concepts as nodes with associated images and detailed textual descriptions. We propose a context-aware multimodal symbol grounding approach to align concept-image and concept-description pairs using context information from image-text datasets. Comprising 951K images and 152K concepts, M^2ConceptBase links each concept to an average of 6.27 images and a single description, ensuring comprehensive visual and textual semantics. Human studies confirm more than 95% alignment accuracy, underscoring its quality. Additionally, our experiments demonstrate that M^2ConceptBase significantly enhances VQA model performance on the OK-VQA task. M^2ConceptBase also substantially improves the fine-grained concept understanding capabilities of multimodal large language models through retrieval augmentation in two concept-related tasks, highlighting its value.

Yicong Zhao, Shisong Chen, Jiacheng Zhang et al.

Recent advances in large language models (LLMs) have demonstrated impressive capabilities in code-related tasks, such as code generation and automated program repair. Despite their promising performance, most existing approaches for code repair suffer from high training costs or computationally expensive inference. Retrieval-augmented generation (RAG), with its efficient in-context learning paradigm, offers a more scalable alternative. However, conventional retrieval strategies, which are often based on holistic code-text embeddings, fail to capture the structural intricacies of code, resulting in suboptimal retrieval quality. To address the above limitations, we propose ReCode, a fine-grained retrieval-augmented in-context learning framework designed for accurate and efficient code repair. Specifically, ReCode introduces two key innovations: (1) an algorithm-aware retrieval strategy that narrows the search space using preliminary algorithm type predictions; and (2) a modular dual-encoder architecture that separately processes code and textual inputs, enabling fine-grained semantic matching between input and retrieved contexts. Furthermore, we propose RACodeBench, a new benchmark constructed from real-world user-submitted buggy code, which addresses the limitations of synthetic benchmarks and supports realistic evaluation. Experimental results on RACodeBench and competitive programming datasets demonstrate that ReCode achieves higher repair accuracy with significantly reduced inference cost, highlighting its practical value for real-world code repair scenarios.

Jipeng Cen, Jiaxin Liu, Zhixu Li et al.

While fine-tuned large language models (LLMs) excel in generating grammatically valid SQL in Text-to-SQL parsing, they often struggle to ensure semantic accuracy in queries, leading to user confusion and diminished system usability. To tackle this challenge, we introduce SQLFixAgent, a new consistency-enhanced multi-agent collaborative framework designed for detecting and repairing erroneous SQL. Our framework comprises a core agent, SQLRefiner, alongside two auxiliary agents: SQLReviewer and QueryCrafter. The SQLReviewer agent employs the rubber duck debugging method to identify potential semantic mismatches between SQL and user query. If the error is detected, the QueryCrafter agent generates multiple SQL as candidate repairs using a fine-tuned SQLTool. Subsequently, leveraging similar repair retrieval and failure memory reflection, the SQLRefiner agent selects the most fitting SQL statement from the candidates as the final repair. We evaluated our proposed framework on five Text-to-SQL benchmarks. The experimental results show that our method consistently enhances the performance of the baseline model, specifically achieving an execution accuracy improvement of over 3% on the Bird benchmark. Our framework also has a higher token efficiency compared to other advanced methods, making it more competitive.

Xiaoxuan Zhu, Zhouhong Gu, Sihang Jiang et al.

Online courses have significantly lowered the barrier to accessing education, yet the varying content quality of these videos poses challenges. In this work, we focus on the task of automatically evaluating the quality of video course content. We have constructed a dataset with a substantial collection of video courses and teaching materials. We propose three evaluation principles and design a new evaluation framework, \textit{VCEval}, based on these principles. The task is modeled as a multiple-choice question-answering task, with a language model serving as the evaluator. Our method effectively distinguishes video courses of different content quality and produces a range of interpretable results.

Jiaan Wang, Fandong Meng, Duo Zheng et al.

To adapt text summarization to the multilingual world, previous work proposes multi-lingual summarization (MLS) and cross-lingual summarization (CLS). However, these two tasks have been studied separately due to the different definitions, which limits the compatible and systematic research on both of them. In this paper, we aim to unify MLS and CLS into a more general setting, i.e., many-to-many summarization (M2MS), where a single model could process documents in any language and generate their summaries also in any language. As the first step towards M2MS, we conduct preliminary studies to show that M2MS can better transfer task knowledge across different languages than MLS and CLS. Furthermore, we propose Pisces, a pre-trained M2MS model that learns language modeling, cross-lingual ability and summarization ability via three-stage pre-training. Experimental results indicate that our Pisces significantly outperforms the state-of-the-art baselines, especially in the zero-shot directions, where there is no training data from the source-language documents to the target-language summaries.

Xiangru Zhu, Zhixu Li, Xiaodan Wang et al.

Recent years have witnessed the resurgence of knowledge engineering which is featured by the fast growth of knowledge graphs. However, most of existing knowledge graphs are represented with pure symbols, which hurts the machine's capability to understand the real world. The multi-modalization of knowledge graphs is an inevitable key step towards the realization of human-level machine intelligence. The results of this endeavor are Multi-modal Knowledge Graphs (MMKGs). In this survey on MMKGs constructed by texts and images, we first give definitions of MMKGs, followed with the preliminaries on multi-modal tasks and techniques. We then systematically review the challenges, progresses and opportunities on the construction and application of MMKGs respectively, with detailed analyses of the strength and weakness of different solutions. We finalize this survey with open research problems relevant to MMKGs.

Jiaan Wang, Beiqi Zou, Zhixu Li et al.

Story ending generation is an interesting and challenging task, which aims to generate a coherent and reasonable ending given a story context. The key challenges of the task lie in how to comprehend the story context sufficiently and handle the implicit knowledge behind story clues effectively, which are still under-explored by previous work. In this paper, we propose a Story Heterogeneous Graph Network (SHGN) to explicitly model both the information of story context at different granularity levels and the multi-grained interactive relations among them. In detail, we consider commonsense knowledge, words and sentences as three types of nodes. To aggregate non-local information, a global node is also introduced. Given this heterogeneous graph network, the node representations are updated through graph propagation, which adequately utilizes commonsense knowledge to facilitate story comprehension. Moreover, we design two auxiliary tasks to implicitly capture the sentiment trend and key events lie in the context. The auxiliary tasks are jointly optimized with the primary story ending generation task in a multi-task learning strategy. Extensive experiments on the ROCStories Corpus show that the developed model achieves new state-of-the-art performances. Human study further demonstrates that our model generates more reasonable story endings.

Jiaan Wang, Zhixu Li, Tingyi Zhang et al.

Sports game summarization aims at generating sports news from live commentaries. However, existing datasets are all constructed through automated collection and cleaning processes, resulting in a lot of noise. Besides, current works neglect the knowledge gap between live commentaries and sports news, which limits the performance of sports game summarization. In this paper, we introduce K-SportsSum, a new dataset with two characteristics: (1) K-SportsSum collects a large amount of data from massive games. It has 7,854 commentary-news pairs. To improve the quality, K-SportsSum employs a manual cleaning process; (2) Different from existing datasets, to narrow the knowledge gap, K-SportsSum further provides a large-scale knowledge corpus that contains the information of 523 sports teams and 14,724 sports players. Additionally, we also introduce a knowledge-enhanced summarizer that utilizes both live commentaries and the knowledge to generate sports news. Extensive experiments on K-SportsSum and SportsSum datasets show that our model achieves new state-of-the-art performances. Qualitative analysis and human study further verify that our model generates more informative sports news.

Jiaan Wang, Zhixu Li, Qiang Yang et al.

Sports game summarization aims to generate news articles from live text commentaries. A recent state-of-the-art work, SportsSum, not only constructs a large benchmark dataset, but also proposes a two-step framework. Despite its great contributions, the work has three main drawbacks: 1) the noise existed in SportsSum dataset degrades the summarization performance; 2) the neglect of lexical overlap between news and commentaries results in low-quality pseudo-labeling algorithm; 3) the usage of directly concatenating rewritten sentences to form news limits its practicability. In this paper, we publish a new benchmark dataset SportsSum2.0, together with a modified summarization framework. In particular, to obtain a clean dataset, we employ crowd workers to manually clean the original dataset. Moreover, the degree of lexical overlap is incorporated into the generation of pseudo labels. Further, we introduce a reranker-enhanced summarizer to take into account the fluency and expressiveness of the summarized news. Extensive experiments show that our model outperforms the state-of-the-art baseline.

Jiaan Wang, Zhixu Li, Binbin Gu et al.

We discuss a novel task, Chorus Recognition, which could potentially benefit downstream tasks such as song search and music summarization. Different from the existing tasks such as music summarization or lyrics summarization relying on single-modal information, this paper models chorus recognition as a multi-modal one by utilizing both the lyrics and the tune information of songs. We propose a multi-modal Chorus Recognition model that considers diverse features. Besides, we also create and publish the first Chorus Recognition dataset containing 627 songs for public use. Our empirical study performed on the dataset demonstrates that our approach outperforms several baselines in chorus recognition. In addition, our approach also helps to improve the accuracy of its downstream task - song search by more than 10.6%.

Hongzhi Yin, Qinyong Wang, Kai Zheng et al.

It has been an important task for recommender systems to suggest satisfying activities to a group of users in people's daily social life. The major challenge in this task is how to aggregate personal preferences of group members to infer the decision of a group. Conventional group recommendation methods applied a predefined strategy for preference aggregation. However, these static strategies are too simple to model the real and complex process of group decision-making, especially for occasional groups which are formed ad-hoc. Moreover, group members should have non-uniform influences or weights in a group, and the weight of a user can be varied in different groups. Therefore, an ideal group recommender system should be able to accurately learn not only users' personal preferences but also the preference aggregation strategy from data. In this paper, we propose a novel end-to-end group recommender system named CAGR (short for Centrality Aware Group Recommender"), which takes Bipartite Graph Embedding Model (BGEM), the self-attention mechanism and Graph Convolutional Networks (GCNs) as basic building blocks to learn group and user representations in a unified way. Specifically, we first extend BGEM to model group-item interactions, and then in order to overcome the limitation and sparsity of the interaction data generated by occasional groups, we propose a self-attentive mechanism to represent groups based on the group members. In addition, to overcome the sparsity issue of user-item interaction data, we leverage the user social networks to enhance user representation learning, obtaining centrality-aware user representations. We create three large-scale benchmark datasets and conduct extensive experiments on them. The experimental results show the superiority of our proposed CAGR by comparing it with state-of-the-art group recommender models.

Pengpeng Zhao, Haifeng Zhu, Yanchi Liu et al.

Next Point-of-Interest (POI) recommendation is of great value for both location-based service providers and users. Recently Recurrent Neural Networks (RNNs) have been proved to be effective on sequential recommendation tasks. However, existing RNN solutions rarely consider the spatio-temporal intervals between neighbor check-ins, which are essential for modeling user check-in behaviors in next POI recommendation. In this paper, we propose a new variant of LSTM, named STLSTM, which implements time gates and distance gates into LSTM to capture the spatio-temporal relation between successive check-ins. Specifically, one-time gate and one distance gate are designed to control short-term interest update, and another time gate and distance gate are designed to control long-term interest update. Furthermore, to reduce the number of parameters and improve efficiency, we further integrate coupled input and forget gates with our proposed model. Finally, we evaluate the proposed model using four real-world datasets from various location-based social networks. Our experimental results show that our model significantly outperforms the state-of-the-art approaches for next POI recommendation.