COLING 2024 论文列表

Implicit Discourse Relation Recognition (IDRR), which is the task of recognizing the semantic relation between given text spans that do not contain overt clues, is a long-standing and challenging problem. In particular, the paucity of training data for some error-prone discourse relations makes the problem even more challenging. To address this issue, we propose a method of generating synthetic data for IDRR using a large language model. The proposed method is summarized as two folds: extraction of confusing discourse relation pairs based on false negative rate and synthesis of data focused on the confusion. The key points of our proposed method are utilizing a confusion matrix and adopting two-stage prompting to obtain effective synthetic data. According to the proposed method, we generated synthetic data several times larger than training examples for some error-prone discourse relations and incorporated it into training. As a result of experiments, we achieved state-of-the-art macro-F1 performance thanks to the synthetic data without sacrificing micro-F1 performance and demonstrated its positive effects especially on recognizing some infrequent discourse relations.

With the increasing availability of multimodal content on social media, consisting primarily of text and images, multimodal named entity recognition (MNER) has gained a wide-spread attention. A fundamental challenge of MNER lies in effectively aligning different modalities. However, the majority of current approaches rely on word-based sequence labeling framework and align the image and text at inconsistent semantic levels (whole image-words or regions-words). This misalignment may lead to inferior entity recognition performance. To address this issue, we propose an effective span-based method, named SMNER, which achieves a more consistent multimodal alignment from the perspectives of information-theoretic and cross-modal interaction, respectively. Specifically, we first introduce a cross-modal information bottleneck module for the global-level multimodal alignment (whole image-whole text). This module aims to encourage the semantic distribution of the image to be closer to the semantic distribution of the text, which can enable the filtering out of visual noise. Next, we introduce a cross-modal attention module for the local-level multimodal alignment (regions-spans), which captures the correlations between regions in the image and spans in the text, enabling a more precise alignment of the two modalities. Extensive ex- periments conducted on two benchmark datasets demonstrate that SMNER outperforms the state-of-the-art baselines.

Data availability is crucial for advancing artificial intelligence applications, including voice-based technologies. As content creation, particularly in social media, experiences increasing demand, translation and text-to-speech (TTS) technologies have become essential tools. Notably, the performance of these TTS technologies is highly dependent on the quality of the training data, emphasizing the mutual dependence of data availability and technological progress. This paper introduces an end-to-end tool to generate high-quality datasets for text-to-speech (TTS) models to address this critical need for high-quality data. The contributions of this work are manifold and include: the integration of language-specific phoneme distribution into sample selection, automation of the recording process, automated and human-in-the-loop quality assurance of recordings, and processing of recordings to meet specified formats. The proposed application aims to streamline the dataset creation process for TTS models through these features, thereby facilitating advancements in voice-based technologies.

It is well known that left conjuncts tend to be shorter in English coordinate structures. On the basis of Penn Treebank, Przepiórkowski and Woźniak 2023 (in ACL 2023 proceedings) show that this tendency depends on the difference between lengths of conjuncts: the larger the difference, the stronger the tendency for the shorter conjunct to occur on the left. However, this dynamics is observed only when the governor of the coordinate structure is on the left of the coordination (e.g., “Bring apples and oranges!”) or when it is absent (e.g., “Come and sing!”), and not when it is on the right (e.g., “Apples and oranges fell”). Given the principle of Dependency Length Minimization, this turns out to provide an argument for the symmetric structure of coordination. We replicate and sharpen this result on the basis of a much larger dataset: parts of the COCA corpus parsed with Stanza. We also investigate the dependence of this result on the assumed unit of length (word vs. character) and on genre.

The task of financial analysis primarily encompasses two key areas: stock trend prediction and the corresponding financial question answering. Currently, machine learning and deep learning algorithms (ML&DL) have been widely applied for stock trend predictions, leading to significant progress. However, these methods fail to provide reasons for predictions, lacking interpretability and reasoning processes. Also, they can not integrate textual information such as financial news or reports. Meanwhile, large language models (LLM) have remarkable textual understanding and generation ability. But due to the scarcity of financial training datasets and limited integration with real-time knowledge, LLM still suffer from hallucinations and unable to keep up with the latest information. To tackle these challenges, we first release AlphaFin datasets, combining traditional research datasets, real-time financial data, and handwritten chain-of-thought (CoT) data. It has positive impact on training LLM for completing financial analysis. We then use AlphaFin datasets to benchmark a state-of-the-art method, called Stock-Chain, for effectively tackling the financial analysis task, which integrates retrieval-augmented generation (RAG) techniques. Extensive experiments are conducted to demonstrate the effectiveness of our framework on financial analysis.

Abstractive summarization commonly suffers from exposure bias caused by supervised teacher-force learning, that a model predicts the next token conditioned on the accurate pre-context during training while on its preceding outputs at inference. Existing solutions bridge this gap through un- or semi-supervised holistic learning yet still leave the risk of error accumulation while generating a summary. In this paper, we attribute this problem to the limitation of unidirectional autoregressive text generation and introduce post-processing steps to alleviate it. Specifically, we reformat abstractive summarization to sequential generation and revision (SeGRe), i.e., a model in the revision phase re-inputs the generated summary and refines it by contrasting it with the source document. This provides the model additional opportunities to assess the flawed summary from a global view and thereby modify inappropriate expressions. Moreover, we train the SeGRe model with a regularized minimum-risk policy to ensure effective generation and revision. A lot of comparative experiments are implemented on two well-known datasets, exhibiting the new or matched state-of-the-art performance of SeGRe.

The visual question localized-answering (VQLA) system has garnered increasing attention due to its potential as a knowledgeable assistant in surgical education. Apart from providing text-based answers, VQLA can also pinpoint the specific region of interest for better surgical scene understanding. Although recent Transformer-based models for VQLA have obtained promising results, they (1) conduct vanilla text-to-image cross attention, leading to unidirectional and coarse-grained alignment; (2) ignore exploiting the semantics of answers to further boost performance. In this paper, we propose a novel model termed OTAS, which first introduces optimal transport to achieve bidirectional and fine-grained alignment between images and questions, enabling more precise localization. Besides, OTAS incorporates a set of learnable candidate answer embeddings to query the probability of each answer class for a given image-question pair. Through Transformer attention, the candidate answer embeddings interact with the fused features of the image-question pair to make the answer decision. Extensive experiments on two widely-used benchmark datasets demonstrate the superiority of our model over state-of-the-art methods.

This paper presents a merged collection of entity and coreference annotated data grounded in the Universal Dependencies (UD) treebanks for the two written forms of Norwegian: Bokmål and Nynorsk. The aligned and converted corpora are the Norwegian Named Entities (NorNE) and Norwegian Anaphora Resolution Corpus (NARC). While NorNE is aligned with an older version of the treebank, NARC is misaligned and requires extensive transformation from the original annotations to the UD structure and CoNLL-U format. Here, we demonstrate the conversion and alignment processes, along with an analysis of discovered issues and errors in the data, some of which include data split overlaps in the original treebank. These procedures and the developed system may prove helpful for future work on processing and aligning data from universal dependencies. The merged corpora comprise the first Norwegian UD treebank enriched with named entities and coreference information, supporting the standardized format for the CorefUD initiative.

The advent of scalable deep models and large datasets has improved the performance of Neural Machine Translation (NMT). Knowledge Distillation (KD) enhances efficiency by transferring knowledge from a teacher model to a more compact student model. However, KD approaches to Transformer architecture often rely on heuristics, particularly when deciding which teacher layers to distill from. In this paper, we introduce the “Align-to-Distill” (A2D) strategy, designed to address the feature mapping problem by adaptively aligning student attention heads with their teacher counterparts during training. The Attention Alignment Module (AAM) in A2D performs a dense head-by-head comparison between student and teacher attention heads across layers, turning the combinatorial mapping heuristics into a learning problem. Our experiments show the efficacy of A2D, demonstrating gains of up to +3.61 and +0.63 BLEU points for WMT-2022 De→Dsb and WMT-2014 En→De, respectively, compared to Transformer baselines.The code and data are available at https://github.com/ncsoft/Align-to-Distill.

Recent advancements in Chain-of-Thought prompting have facilitated significant breakthroughs for Large Language Models (LLMs) in complex reasoning tasks. Current research enhances the reasoning performance of LLMs by sampling multiple reasoning chains and ensembling based on the answer frequency. However, this approach fails in scenarios where the correct answers are in the minority. We identify this as a primary factor constraining the reasoning capabilities of LLMs, a limitation that cannot be resolved solely based on the predicted answers. To address this shortcoming, we introduce a hierarchical reasoning aggregation framework AoR (Aggregation of Reasoning), which selects answers based on the evaluation of reasoning chains. Additionally, AoR incorporates dynamic sampling, adjusting the number of reasoning chains in accordance with the complexity of the task. Experimental results on a series of complex reasoning tasks show that AoR outperforms prominent ensemble methods. Further analysis reveals that AoR not only adapts various LLMs but also achieves a superior performance ceiling when compared to current methods.

In this paper we present a series of experiments towards POS tagging Corsican, a less-resourced language spoken in Corsica and linguistically related to Italian. The first contribution is Corsican-POS, the first gold standard POS-tagged corpus for Corsica, composed of 500 sentences manually annotated with the Universal POS tagset. Our second contribution is a set of experiments and evaluation of POS tagging models which starts with a baseline model for Italian and is aimed at finding the best training configuration, namely in terms of the size and combination strategy of the existing raw and annotated resources. These experiments result in (i) the first POS tagger for Corsican, reaching an accuracy of 93.38%, (ii) a quantification of the gain provided by the use of each available resource. We find that the optimal configuration uses Italian word embeddings further specialized with Corsican embeddings and trained on the largest gold corpus for Corsican available so far.

Language change has been the subject of numerous studies in linguistics. However, due to the dynamic and complex nature of this phenomenon, and to the difficulty of obtaining extensive real data of language in use, some of its aspects remain obscure. In recent years, nonetheless, research has used computational modeling to simulate features related to variation, change, propagation, and evolution of languages in speech communities, finding compelling results. In this article, agent-based modeling and simulation is used to study language change. Drawing on previous studies, a speech community was modeled using Zachary’s karate club network, a well-established small-world network model in the field of complex systems. Idiolects were assigned through numerical values for each agent. The results demonstrate that the centrality of each agent in the network, interpreted as social prestige, appears to be a factor influencing change. Additionally, the nature of idiolects also seems to impact the spread of linguistic variants in the language change process. These findings complement the theoretical understanding of the language change phenomenon with new simulation data and provide new avenues for research.

This paper introduces a novel problem of automated question generation for courtroom examinations, CourtQG. While question generation has been studied in domains such as educational testing and product description, CourtQG poses several unique challenges owing to its non-cooperative and agenda-driven nature. Specifically, not only the generated questions need to be relevant to the case and underlying context, they also have to achieve certain objectives such as challenging the opponent’s arguments and/or revealing potential inconsistencies in their answers. We propose to leverage large language models (LLM) for CourtQG by fine-tuning them on two auxiliary tasks, agenda explanation (i.e., uncovering the underlying intents) and question type prediction. We additionally propose cold-start generation of questions from background documents without relying on examination history. We construct a dataset to evaluate our proposed method and show that it generates better questions according to standard metrics when compared to several baselines.

Topic segmentation and outline generation strive to divide a document into coherent topic sections and generate corresponding subheadings, unveiling the discourse topic structure of a document. Compared with sentence-level topic structure, the paragraph-level topic structure can quickly grasp and understand the overall context of the document from a higher level, benefitting many downstream tasks such as summarization, discourse parsing, and information retrieval. However, the lack of large-scale, high-quality Chinese paragraph-level topic structure corpora restrained relative research and applications. To fill this gap, we build the Chinese paragraph-level topic representation, corpus, and benchmark in this paper. Firstly, we propose a hierarchical paragraph-level topic structure representation with three layers to guide the corpus construction. Then, we employ a two-stage man-machine collaborative annotation method to construct the largest Chinese Paragraph-level Topic Structure corpus (CPTS), achieving high quality. We also build several strong baselines, including ChatGPT, to validate the computability of CPTS on two fundamental tasks (topic segmentation and outline generation) and preliminarily verified its usefulness for the downstream task (discourse parsing).

Semi-supervised learning that leverages synthetic data for training has been widely adopted for developing automatic post-editing (APE) models due to the lack of training data. With this aim, we focus on data-synthesis methods to create high-quality synthetic data. Given that APE takes as input a machine-translation result that might include errors, we present a data-synthesis method by which the resulting synthetic data mimic the translation errors found in actual data. We introduce a noising-based data-synthesis method by adapting the masked language model approach, generating a noisy text from a clean text by infilling masked tokens with erroneous tokens. Moreover, we propose selective corpus interleaving that combines two separate synthetic datasets by taking only the advantageous samples to enhance the quality of the synthetic data further. Experimental results show that using the synthetic data created by our approach results in significantly better APE performance than other synthetic data created by existing methods.

In the realm of conversational dynamics, individual idiosyncrasies challenge the suitability of a one-size-fits-all approach for dialogue agent responses. Prior studies often assumed the speaker’s persona’s immediate availability, a premise not universally applicable. To address this gap, we explore the Speaker Profiling in Conversations (SPC) task, aiming to synthesize persona attributes for each dialogue participant. SPC comprises three core subtasks: persona discovery, persona-type identification, and persona-value extraction. The first subtask identifies persona-related utterances, the second classifies specific attributes, and the third extracts precise values for the persona. To confront this multifaceted challenge, we’ve diligently compiled SPICE, an annotated dataset, underpinning our thorough evaluation of diverse baseline models. Additionally, we benchmark these findings against our innovative neural model, SPOT, presenting an exhaustive analysis encompassing a nuanced assessment of quantitative and qualitative merits and limitations.

Traditional non-simultaneous Sign Language Translation (SLT) methods, while effective for pre-recorded videos, face challenges in real-time scenarios due to inherent inference delays. The emerging field of simultaneous SLT aims to address this issue by progressively translating incrementally received sign video. However, the sole existing work in simultaneous SLT adopts a fixed gloss-based policy, which suffer from limitations in boundary prediction and contextual comprehension. In this paper, we delve deeper into this area and propose an adaptive policy for simultaneous SLT. Our approach introduces the concept of “confident translation length”, denoting maximum accurate translation achievable from current input. An estimator measures this length for streaming sign video, enabling the model to make informed decisions on whether to wait for more input or proceed with translation. To train the estimator, we construct a training data of confident translation length based on the longest common prefix between translations of partial and complete inputs. Furthermore, we incorporate adaptive training, utilizing pseudo prefix pairs, to refine the offline translation model for optimal performance in simultaneous scenarios. Experimental results on PHOENIX2014T and CSL-Daily demonstrate the superiority of our adaptive policy over existing methods, particularly excelling in situations requiring extremely low latency.

Sentence representation learning is a fundamental task in NLP. Existing methods use contrastive learning (CL) to learn effective sentence representations, which benefit from high-quality contrastive data but require extensive human annotation. Large language models (LLMs) like ChatGPT and GPT4 can automatically generate such data. However, this alternative strategy also encounters challenges: 1) obtaining high-quality generated data from small-parameter LLMs is difficult, and 2) inefficient utilization of the generated data. To address these challenges, we propose a novel adaptive reinforcement tuning (ART) framework. Specifically, to address the first challenge, we introduce a reinforcement learning approach for fine-tuning small-parameter LLMs, enabling the generation of high-quality hard contrastive data without human feedback. To address the second challenge, we propose an adaptive iterative framework to guide the small-parameter LLMs to generate progressively harder samples through multiple iterations, thereby maximizing the utility of generated data. Experiments conducted on seven semantic text similarity tasks demonstrate that the sentence representation models trained using the synthetic data generated by our proposed method achieve state-of-the-art performance. Our code is available at https://github.com/WuNein/AdaptCL.

The fine-tuning paradigm has been widely adopted to train neural models tailored for specific tasks. However, the recent upsurge of Large Language Models (LLMs), characterized by billions of parameters, has introduced profound computational challenges to the fine-tuning process. This has fueled intensive research on Parameter-Efficient Fine-Tuning (PEFT) techniques, usually involving the training of a selective subset of the original model parameters. One of the most used approaches is Adapters, which add trainable lightweight layers to the existing pretrained weights. Within this context, we propose AdaKron, an Adapter-based fine-tuning with the Kronecker product. In particular, we leverage the Kronecker product to combine the output of two small networks, resulting in a final vector whose dimension is the product of the dimensions of the individual outputs, allowing us to train only 0.55% of the model’s original parameters. We evaluate AdaKron performing a series of experiments on the General Language Understanding Evaluation (GLUE) benchmark, achieving results in the same ballpark as recent state-of-the-art PEFT methods, despite training fewer parameters.