COLING 2024 Papers

Reinforcement learning from human feedback (RLHF) is a crucial technique in aligning large language models (LLMs) with human preferences, ensuring these LLMs behave in beneficial and comprehensible ways to users. However, a longstanding challenge in human alignment techniques based on reinforcement learning lies in their inherent complexity and difficulty in training. To address this challenge, we present a simple yet effective Contrastive Learning Framework for Human Alignment (CLHA) to align LLMs with human preferences directly. CLHA employs a novel rescoring strategy to evaluate the noise within the data by considering its inherent quality and dynamically adjusting the training process. Simultaneously, CLHA utilizes pairwise contrastive loss and adaptive supervised fine-tuning loss to adaptively modify the likelihood of generating responses, ensuring enhanced alignment with human preferences. Using advanced methods, CLHA surpasses other algorithms, showcasing superior performance in terms of reward model scores, automatic evaluations, and human assessments on the widely used “Helpful and Harmless” dataset.

In an era where rumors can propagate rapidly across social media platforms such as Twitter and Weibo, automatic rumor detection has garnered considerable attention from both academia and industry. Existing multimodal rumor detection models often overlook the intricacies of sample difficulty, e.g., text-level difficulty, image-level difficulty, and multimodal-level difficulty, as well as their order when training. Inspired by the concept of curriculum learning, we propose the Curriculum Learning and Fine-grained Fusion-driven multimodal Rumor Detection (CLFFRD) framework, which employs curriculum learning to automatically select and train samples according to their difficulty at different training stages. Furthermore, we introduce a fine-grained fusion strategy that unifies entities from text and objects from images, enhancing their semantic cohesion. We also propose a novel data augmentation method that utilizes linear interpolation between textual and visual modalities to generate diverse data. Additionally, our approach incorporates deep fusion for both intra-modality (e.g., text entities and image objects) and inter-modality (e.g., CLIP and social graph) features. Extensive experimental results demonstrate that CLFFRD outperforms state-of-the-art models on both English and Chinese benchmark datasets for rumor detection in social media.

The integration of learning and reasoning is high on the research agenda in AI. Nevertheless, there is only a little attention to using existing background knowledge for reasoning about partially observed scenes to answer questions about the scene. Yet, we as humans use such knowledge frequently to infer plausible answers to visual questions (by eliminating all inconsistent ones). Such knowledge often comes in the form of constraints about objects and it tends to be highly domain or environment specific. We contribute a novel benchmark called CLEVR-POC for reasoning-intensive visual question answering (VQA) in partially observable environments under constraints. In CLEVR-POC, knowledge in the form of logical constraints needs to be leveraged in order to generate plausible answers to questions about a hidden object in a given partial scene. For instance, if one has the knowledge that all cups are colored either red, green or blue and that there is only one green cup, it becomes possible to deduce the color of an occluded cup as either red or blue, provided that all other cups, including the green one, are observed. Through experiments we observe that the performance of pre-trained vision language models like CLIP (approx. 22%) and a large language model (LLM) like GPT-4 (approx. 46%) on CLEVR-POC are not satisfactory, ascertaining the necessity for frameworks that can handle reasoning-intensive tasks where environment-specific background knowledge is available and crucial. Furthermore, our demonstration illustrates that a neuro-symbolic model, which integrates an LLM like GPT-4 with a visual perception network and a formal logical reasoner, exhibits exceptional performance on CLEVR-POC.

We introduce CLAUSE-ATLAS, a resource of XIX and XX century English novels annotated automatically. This corpus, which contains 41,715 labeled clauses, allows to study stories as sequences of eventive, subjective and contextual information. We use it to investigate if recent large language models, in particular gpt-3.5-turbo with 16k tokens of context, constitute promising tools to annotate large amounts of data for literary studies (we show that this is the case). Moreover, by analyzing the annotations so collected, we find that our clause-based approach to literature captures structural patterns within books, as well as qualitative differences between them.

This paper presents a collection of highly comparable web corpora of Slovenian, Croatian, Bosnian, Montenegrin, Serbian, Macedonian, and Bulgarian, covering thereby the whole spectrum of official languages in the South Slavic language space. The collection of these corpora comprises a total of 13 billion tokens of texts from 26 million documents. The comparability of the corpora is ensured by a comparable crawling setup and the usage of identical crawling and post-processing technology. All the corpora were linguistically annotated with the state-of-the-art CLASSLA-Stanza linguistic processing pipeline, and enriched with document-level genre information via the Transformer-based multilingual X-GENRE classifier, which further enhances comparability at the level of linguistic annotation and metadata enrichment. The genre-focused analysis of the resulting corpora shows a rather consistent distribution of genres throughout the seven corpora, with variations in the most prominent genre categories being well-explained by the economic strength of each language community. A comparison of the distribution of genre categories across the corpora indicates that web corpora from less developed countries primarily consist of news articles. Conversely, web corpora from economically more developed countries exhibit a smaller proportion of news content, with a greater presence of promotional and opinionated texts.

Natural Language Processing (NLP) plays a pivotal role in the realm of Digital Humanities (DH) and serves as the cornerstone for advancing the structural analysis of historical and cultural heritage texts. This is particularly true for the domains of named entity recognition (NER) and relation extraction (RE). In our commitment to expediting ancient history and culture, we present the “Chinese Historical Information Extraction Corpus”(CHisIEC). CHisIEC is a meticulously curated dataset designed to develop and evaluate NER and RE tasks, offering a resource to facilitate research in the field. Spanning a remarkable historical timeline encompassing data from 13 dynasties spanning over 1830 years, CHisIEC epitomizes the extensive temporal range and text heterogeneity inherent in Chinese historical documents. The dataset encompasses four distinct entity types and twelve relation types, resulting in a meticulously labeled dataset comprising 14,194 entities and 8,609 relations. To establish the robustness and versatility of our dataset, we have undertaken comprehensive experimentation involving models of various sizes and paradigms. Additionally, we have evaluated the capabilities of Large Language Models (LLMs) in the context of tasks related to ancient Chinese history. The dataset and code are available at https://github.com/tangxuemei1995/CHisIEC.

Existing studies of naturally occurring language-in-interaction have largely focused on the two ends of the developmental spectrum, i.e., early childhood and adulthood, leaving a gap in our knowledge about how development unfolds, especially across middle childhood. The current work contributes to filling this gap by introducing CHICA (for Child Interpersonal Communication Analysis), a developmental corpus of child-caregiver conversations at home, involving groups of French-speaking children aged 7, 9, and 11 years old. Each dyad was recorded twice: once in a face-to-face setting and once using computer-mediated video calls. For the face-to-face settings, we capitalized on recent advances in mobile, lightweight eye-tracking and head motion detection technology to optimize the naturalness of the recordings, allowing us to obtain both precise and ecologically valid data. Further, we mitigated the challenges of manual annotation by relying – to the extent possible – on automatic tools in speech processing and computer vision. Finally, to demonstrate the richness of this corpus for the study of child communicative development, we provide preliminary analyses comparing several measures of child-caregiver conversational dynamics across developmental age, modality, and communicative medium. We hope the current corpus will allow new discoveries into the properties and mechanisms of multimodal communicative development across middle childhood.

Emotion Recognition in Conversation (ERC) has attracted increasing attention due to its wide applications in public opinion analysis, empathetic conversation generation, and so on. However, ERC research suffers from the problems of data imbalance and the presence of similar linguistic expressions for different emotions. These issues can result in limited learning for minority emotions, biased predictions for common emotions, and the misclassification of different emotions with similar linguistic expressions. To alleviate these problems, we propose a Contrast-Enhanced Prompt-Tuning (CEPT) framework for ERC. We transform the ERC task into a Masked Language Modeling (MLM) generation task and generate the emotion for each utterance in the conversation based on the prompt-tuning of the Pre-trained Language Model (PLM), where a novel mixed prompt template and a label mapping strategy are introduced for better context and emotion feature modeling. Moreover, Supervised Contrastive Learning (SCL) is employed to help the PLM mine more information from the labels and learn a more discriminative representation space for utterances with different emotions. We conduct extensive experiments and the results demonstrate that CEPT outperforms the state-of-the-art methods on all three benchmark datasets and excels in recognizing minority emotions.

The Video-Grounded Dialogue generation (VDG) is a challenging task requiring a comprehensive understanding of the multi-modal information to produce a pertinent response. However, VDG models may rely on dataset bias as a shortcut and fail to learn the multi-modal knowledge from both video and audio. Counterfactual reasoning is an effective method that can estimate and eliminate bias on some special aspects of classification tasks. However, conventional counterfactual reasoning cannot be applied to VDG tasks directly due to the BPE algorithm. In this paper, we reformulate the counterfactual reasoning from the information entropy perspective and extend it from the classification task to the generative task, which can effectively reduce the question-related bias in the auto-regressive generation task. We design CE-VDG to demonstrate the effectiveness in bias elimination of the reformulated counterfactual reasoning by using the proposed counterfactual entropy as an external loss. Extensive experiment results on two popular VDG datasets show the superiority of CE-VDG over the existing baseline method, demonstrating the effective debiasing capability in our model considering counterfactual entropy.

The recent advancements in artificial intelligence highlight the potential of language models in psychological health support. While models trained on data from mental health service platform have achieved preliminary success, challenges persist in areas such as data scarcity, quality, and ensuring a solid foundation in psychological techniques. To address these challenges, this study introduces a novel approach to enhance the precision and efficacy of psychological support through large language models. Specifically, we design a specific prompt derived from principles of Cognitive Behavioral Therapy (CBT) and have generated the CBT QA dataset, specifically for Chinese psychological health Q&A based on CBT structured intervention strategies. Unlike previous methods, our dataset emphasizes professional and structured response. Utilizing this dataset, we fine-tuned the large language model, giving birth to CBT-LLM, the large-scale language model specifically designed for Cognitive Behavioral Therapy techniques. Empirical evaluations demonstrate that CBT-LLM excels in generating structured, professional, and highly relevant responses in psychological health support tasks, showcasing its practicality and quality. The model is available on Hugging Face: https://huggingface.co/Hongbin37/CBT-LLM.

Holistically measuring societal biases of large language models is crucial for detecting and reducing ethical risks in highly capable AI models. In this work, we present a Chinese Bias Benchmark dataset that consists of over 100K questions jointly constructed by human experts and generative language models, covering stereotypes and societal biases in 14 social dimensions related to Chinese culture and values. The curation process contains 4 essential steps: bias identification, ambiguous context generation, AI-assisted disambiguous context generation, and manual review and recomposition. The testing instances in the dataset are automatically derived from 3K+ high-quality templates manually authored with stringent quality control. The dataset exhibits wide coverage and high diversity. Extensive experiments demonstrate the effectiveness of the dataset in evaluating model bias, with all 12 publicly available Chinese large language models exhibiting strong bias in certain categories. Additionally, we observe from our experiments that fine-tuned models could, to a certain extent, heed instructions and avoid generating harmful outputs, in the way of “moral self-correction”. Our dataset is available at https://anonymous.4open.science/r/CBBQ-B860/.

End-to-end automatic speech recognition (ASR) systems often struggle to recognize rare name entities, such as personal names, organizations and terminologies that are not frequently encountered in the training data. This paper presents Contextual Biasing Whisper (CB-Whisper), a novel ASR system based on OpenAI’s Whisper model that can recognize user-defined name entities by performing open-vocabulary keyword-spotting (KWS) before the decoder. The KWS module leverages text-to-speech (TTS) techniques and a convolutional neural network (CNN) classifier to match the features between the entities and the utterances. To integrate the recognized entities into the Whipser decoder and avoid hallucinations, we carefully crafted multiple prompts with spoken form hints. Experiments show that the KWS module based on Whisper encoder’s features can recognize unseen user-defined keywords effectively. More importantly, the proposed CB-Whisper substantially improves the mixed-error-rate (MER) and entity recall compared to the original Whisper model on three internal datasets and two publicly available datasets including Aishell and ACL datasets that cover English-only, Chinese-only, and code-switching scenarios.

Writing a scientific article is a challenging task as it is a highly codified and specific genre, consequently proficiency in written communication is essential for effectively conveying research findings and ideas. In this article, we propose an original textual resource on the revision step of the writing process of scientific articles. This new dataset, called CASIMIR, contains the multiple revised versions of 15,646 scientific articles from OpenReview, along with their peer reviews. Pairs of consecutive versions of an article are aligned at sentence-level while keeping paragraph location information as metadata for supporting future revision studies at the discourse level. Each pair of revised sentences is enriched with automatically extracted edits and associated revision intention. To assess the initial quality on the dataset, we conducted a qualitative study of several state-of-the-art text revision approaches and compared various evaluation metrics. Our experiments led us to question the relevance of the current evaluation methods for the text revision task.

Joint entity and relation extraction is the fundamental task of information extraction, consisting of two subtasks: named entity recognition and relation extraction. However, most existing joint extraction methods suffer from issues of feature confusion or inadequate interaction between the two subtasks. Addressing these challenges, in this work, we propose a Co-Attention network for joint entity and Relation Extraction (CARE). Our approach includes adopting a parallel encoding strategy to learn separate representations for each subtask, aiming to avoid feature overlap or confusion. At the core of our approach is the co-attention module that captures two-way interaction between the two subtasks, allowing the model to leverage entity information for relation prediction and vice versa, thus promoting mutual enhancement. Through extensive experiments on three benchmark datasets for joint entity and relation extraction (NYT, WebNLG, and SciERC), we demonstrate that our proposed model outperforms existing baseline models. Our code will be available at https://github.com/kwj0x7f/CARE.

Ad text generation is the task of creating compelling text from an advertising asset that describes products or services, such as a landing page. In advertising, diversity plays an important role in enhancing the effectiveness of an ad text, mitigating a phenomenon called “ad fatigue,” where users become disengaged due to repetitive exposure to the same advertisement. Despite numerous efforts in ad text generation, the aspect of diversifying ad texts has received limited attention, particularly in non-English languages like Japanese. To address this, we present CAMERA³, an evaluation dataset for controllable text generation in the advertising domain in Japanese. Our dataset includes 3,980 ad texts written by expert annotators, taking into account various aspects of ad appeals. We make CAMERA³ publicly available, allowing researchers to examine the capabilities of recent NLG models in controllable text generation in a real-world scenario.

Understanding the discussion moves that teachers and students use to engage in classroom discussions is important to support pre-service teacher learning and teacher educators. This work introduces a novel conversational multi-label corpus of teaching transcripts collected from a simulated classroom environment for Conversational Argument Move AnaLysis (CAMAL). The dataset offers various argumentation moves used by pre-service teachers and students in mathematics and science classroom discussions. The dataset includes 165 transcripts from these discussions that pre-service elementary teachers facilitated in a simulated classroom environment of five student avatars. The discussion transcripts were annotated by education assessment experts for nine argumentation moves (aka. intents) used by the pre-service teachers and students during the discussions. In this paper, we describe the dataset, our annotation framework, and the models we employed to detect argumentation moves. Our experiments with state-of-the-art models demonstrate the complexity of the CAMAL task presented in the dataset. The result reveals that models that combined CNN and LSTM structures with speaker ID graphs improved the F1-score of our baseline models to detect speakers’ intents by a large margin. Given the complexity of the CAMAL task, it creates research opportunities for future studies. We share the dataset, the source code, and the annotation framework publicly at http://github.com/uonlp/camal-dataset.

Comparative Question Answering (CompQA) is a Natural Language Processing task that combines Question Answering and Argument Mining approaches to answer subjective comparative questions in an efficient argumentative manner. In this paper, we present an end-to-end (full pipeline) system for answering comparative questions called CAM 2.0 as well as a public leaderboard called CompUGE that unifies the existing datasets under a single easy-to-use evaluation suite. As compared to previous web-form-based CompQA systems, it features question identification, object and aspect labeling, stance classification, and summarization using up-to-date models. We also select the most time- and memory-effective pipeline by comparing separately fine-tuned Transformer Encoder models which show state-of-the-art performance on the subtasks with Generative LLMs in few-shot and LoRA setups. We also conduct a user study for a whole-system evaluation.

We present Component ALignment for Abstract Meaning Representation (Calamr), a novel method for graph alignment that can support summarization and its evaluation. First, our method produces graphs that explain what is summarized through their alignments, which can be used to train graph based summarization learners. Second, although numerous scoring methods have been proposed for abstract meaning representation (AMR) that evaluate semantic similarity, no AMR based summarization metrics exist despite years of work using AMR for this task. Calamr provides alignments on which new scores can be based. The contributions of this work include a) a novel approach to aligning AMR graphs, b) a new summarization based scoring methods for similarity of AMR subgraphs composed of one or more sentences, and c) the entire reusable source code to reproduce our results.

Auxiliary information, such as knowledge graph (KG), has become increasingly crucial in recommender systems. However, the current KG-based recommendation still has some limitations: (1) low link rates between items and KG entities, (2) redundant knowledge in KG. In this paper, we introduce the aspect, which refers to keywords describing item attributes in reviews, to KG-based recommendation, and propose a new model, Collaborative Aspect Graph enhanced Knowledge-based Network (CAGK). Firstly, CAGK builds a Collaborative Aspect Graph (CAG) with user-item interactions, aspects and KG, where aspects can fill most of the sparsity. Secondly, we leverage interactive information and aspect features to generate aspect-aware guidance signals to customize knowledge extraction and eliminate redundant knowledge. Lastly, we utilize low ratings and negative aspect sentiment to capture features of that users dislike to prevent repetitive recommendations of disliked items. Experimental results on two widely used benchmark datasets, Amazon-book and Yelp2018, confirm the superiority of CAGK.