COLING 2024 论文列表

Conversational retrieval refers to an information retrieval system that operates in an iterative and interactive manner, requiring the retrieval of various external resources, such as persona, knowledge, and even response, to effectively engage with the user and successfully complete the dialogue. However, most previous work trained independent retrievers for each specific resource, resulting in sub-optimal performance and low efficiency. Thus, we propose a multi-task framework function as a universal retriever for three dominant retrieval tasks during the conversation: persona selection, knowledge selection, and response selection. To this end, we design a dual-encoder architecture consisting of a context-adaptive dialogue encoder and a candidate encoder, aiming to attention to the relevant context from the long dialogue and retrieve suitable candidates by simply a dot product. Furthermore, we introduce two loss constraints to capture the subtle relationship between dialogue context and different candidates by regarding historically selected candidates as hard negatives. Extensive experiments and analysis establish state-of-the-art retrieval quality both within and outside its training domain, revealing the promising potential and generalization capability of our model to serve as a universal retriever for different candidate selection tasks simultaneously.

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

Recent researches have shown that multi-task instruction tuning after pre-training greatly improves the model’s robustness and transfer ability, which is crucial for building a high-quality dialog system. However, most previous works on multi-task instruction tuning rely heavily on human-defined input format or prompt, which is not optimal in quality and quantity.In this work, we propose to use Task-aware Automatic Prompt generation (TAP) to automatically generate high-quality prompts. Using the high-quality prompts generated, we scale the corpus of the pre-trained conversation model to 122 datasets from 15 dialog-related tasks, resulting in Universal Pre-trained Conversation Model (UniPCM), a powerful foundation model for various conversational tasks and different dialog systems. Extensive experiments have shown that UniPCM is robust to input prompts and capable of various dialog-related tasks. Moreover, UniPCM has strong transfer ability and excels at low resource scenarios, achieving SOTA results on 9 different datasets ranging from task-oriented dialog to open-domain conversation. Furthermore, we are amazed to find that TAP can generate prompts on par with those collected with crowdsourcing.

A transformer model is used in general tasks such as pre-trained language models and specific tasks including machine translation. Such a model mainly relies on positional encodings (PEs) to handle the sequential order of input vectors. There are variations of PEs, such as absolute and relative, and several studies have reported on the superiority of relative PEs. In this paper, we focus on analyzing in which part of a transformer model PEs work and the different characteristics between absolute and relative PEs through a series of experiments. Experimental results indicate that PEs work in both self- and cross-attention blocks in a transformer model, and PEs should be added only to the query and key of an attention mechanism, not to the value. We also found that applying two PEs in combination, a relative PE in the self-attention block and an absolute PE in the cross-attention block, can improve translation quality.

Large language models, like ChatGPT, have shown remarkable capability in many downstream tasks, yet their ability to understand discourse structures of dialogues remains less explored, where it requires higher level capabilities of understanding and reasoning. In this paper, we aim to systematically inspect ChatGPT’s performance in two discourse analysis tasks: topic segmentation and discourse parsing, focusing on its deep semantic understanding of linear and hierarchical discourse structures underlying dialogue. To instruct ChatGPT to complete these tasks, we initially craft a prompt template consisting of the task description, output format, and structured input. Then, we conduct experiments on four popular topic segmentation datasets and two discourse parsing datasets. The experimental results showcase that ChatGPT demonstrates proficiency in identifying topic structures in general-domain conversations yet struggles considerably in specific-domain conversations. We also found that ChatGPT hardly understands rhetorical structures that are more complex than topic structures. Our deeper investigation indicates that ChatGPT can give more reasonable topic structures than human annotations but only linearly parses the hierarchical rhetorical structures. In addition, we delve into the impact of in-context learning (e.g., chain-of-thought) on ChatGPT and conduct the ablation study on various prompt components, which can provide a research foundation for future work. The code is available at https://github.com/yxfanSuda/GPTforDDA.

The behavior and decision making of groups or communities can be dramatically influenced by individuals pushing particular agendas, e.g., to promote or disparage a person or an activity, to call for action, etc.. In the examination of online influence campaigns, particularly those related to important political and social events, scholars often concentrate on identifying the sources responsible for setting and controlling the agenda (e.g., public media). In this article we present a methodology for detecting specific instances of agenda control through social media where annotated data is limited or non-existent. By using a modest corpus of Twitter messages centered on the 2022 French Presidential Elections, we carry out a comprehensive evaluation of various approaches and techniques that can be applied to this problem. Our findings demonstrate that by treating the task as a textual entailment problem, it is possible to overcome the requirement for a large annotated training dataset.

Hate speech detection has become an urgent task with the emergence of huge multimodal harmful content (, memes) on social media platforms. Previous studies mainly focus on complex feature extraction and fusion to learn discriminative information from memes. However, these methods ignore two key points: 1) the misalignment of image and text in memes caused by the modality gap, and 2) the uncertainty between modalities caused by the contribution degree of each modality to hate sentiment. To this end, this paper proposes an uncertainty-aware cross-modal alignment (UCA) framework for modeling the misalignment and uncertainty in multimodal hate speech detection. Specifically, we first utilize the cross-modal feature encoder to capture image and text feature representations in memes. Then, a cross-modal alignment module is applied to reduce semantic gaps between modalities by aligning the feature representations. Next, a cross-modal fusion module is designed to learn semantic interactions between modalities to capture cross-modal correlations, providing complementary features for memes. Finally, a cross-modal uncertainty learning module is proposed, which evaluates the divergence between unimodal feature distributions to to balance unimodal and cross-modal fusion features. Extensive experiments on five publicly available datasets show that the proposed UCA produces a competitive performance compared with the existing multimodal hate speech detection methods.

Named entity recognition as it is traditionally envisioned excludes in practice a significant part of the entities of potential interest for real-word applications: nested, discontinuous, non-named entities. Despite various attempts to broaden their coverage, subsequent annotation schemes have achieved little adoption in the literature and the most restrictive variant of NER remains the default. This is partly due to the complexity of those annotations and their format. In this paper, we introduce a new annotation scheme that offers higher comprehensiveness while preserving simplicity, together with an annotation tool to implement that scheme. We also release the corpus UkraiNER, comprised of 10,000 French sentences in the geopolitical news domain and manually annotated with comprehensive entity recognition. Our baseline experiments on UkraiNER provide a first point of comparison to facilitate future research (82 F1 for comprehensive entity recognition, 87 F1 when focusing on traditional nested NER), as well as various insights on the composition and challenges that this corpus presents for state-of-the-art named entity recognition models.

This paper introduces UQA, a novel dataset for question answering and text comprehension in Urdu, a low-resource language with over 70 million native speakers. UQA is generated by translating the Stanford Question Answering Dataset (SQuAD2.0), a large-scale English QA dataset, using a technique called EATS (Enclose to Anchor, Translate, Seek), which preserves the answer spans in the translated context paragraphs. The paper describes the process of selecting and evaluating the best translation model among two candidates: Google Translator and Seamless M4T. The paper also benchmarks several state-of-the-art multilingual QA models on UQA, including mBERT, XLM-RoBERTa, and mT5, and reports promising results. For XLM-RoBERTa-XL, we have an F1 score of 85.99 and 74.56 EM. UQA is a valuable resource for developing and testing multilingual NLP systems for Urdu and for enhancing the cross-lingual transferability of existing models. Further, the paper demonstrates the effectiveness of EATS for creating high-quality datasets for other languages and domains. The UQA dataset and the code are publicly available at www.github.com/sameearif/UQA

Prior research in Image Machine Translation (IMT) has focused on either translating the source image solely into the target language text or exclusively into the target image. As a result, the former approach lacked the capacity to generate target images, while the latter was insufficient in producing target text. In this paper, we present a Unified Multimodal Text Image Translation (UMTIT) model that not only translates text images into the target language but also generates consistent target images. The UMTIT model consists of two image-text modality conversion steps: the first step converts images to text to recognize the source text and generate translations, while the second step transforms text to images to create target images based on the translations. Due to the limited availability of public datasets, we have constructed two multimodal image translation datasets. Experimental results show that our UMTIT model is versatile enough to handle tasks across multiple modalities and outperforms previous methods. Notably, UMTIT surpasses the state-of-the-art TrOCR in text recognition tasks, achieving a lower Character Error Rate (CER); it also outperforms cascading methods in text translation tasks, obtaining a higher BLEU score; and, most importantly, UMTIT can generate high-quality target text images.

UDMorph provides an infrastructure parallel to that provided by UD for annotated corpus data that follow the UD guidelines, but do not provide dependency relations: a place where new annotated data-sets can be deposited, and existing data-sets can be found and downloaded. It also provides a corpus creation environment to easily create annotated data for additional languages. And it provides a REST and GUI interface to a growing collection taggers with a CoNLL-U output, currently for around 150 different languages.

The Universal Dependencies (UD) project has created an invaluable collection of treebanks with contributions in over 140 languages. However, the UD annotations do not tell the full story. Grammatical constructions that convey meaning through a particular combination of several morphosyntactic elements—for example, interrogative sentences with special markers and/or word orders—are not labeled holistically. We argue for (i) augmenting UD annotations with a ‘UCxn’ annotation layer for such meaning-bearing grammatical constructions, and (ii) approaching this in a typologically informed way so that morphosyntactic strategies can be compared across languages. As a case study, we consider five construction families in ten languages, identifying instances of each construction in UD treebanks through the use of morphosyntactic patterns. In addition to findings regarding these particular constructions, our study yields important insights on methodology for describing and identifying constructions in language-general and language-particular ways, and lays the foundation for future constructional enrichment of UD treebanks.

Dense retrieval (DR) has become a mainstream approach to information seeking, where a system is required to return relevant information to a user query. In real-life applications, typoed queries resulting from the users’ mistyping words or phonetic typing errors exist widely in search behaviors. Current dense retrievers experience a significant drop in retrieval effectiveness when they encounter typoed queries. Therefore, the search system requires the extra introduction of spell-checkers to deal with typos and then applies the DR model to perform robust matching. Herein, we argue that directly conducting the typos correction training would be beneficial to make an end-to-end retriever against misspellings. To this end, we propose a novel approach that can facilitate the incorporation of the spelling correction objective into the DR model using the encoder-decoder architecture. During typos correction training, we also develop a prompt-based augmentation technique to enhance the DR space alignment of the typoed query and its original query. Extensive experiments demonstrate that the effectiveness of our proposed end-to-end retriever significantly outperforms existing typos-aware training approaches and sophisticated training advanced retrievers. Our code is available at https://github.com/striver314/ToCoTR.

In Tokenization and the Noiseless Channel (Zouhar et al., 2023), Rényi efficiency is suggested as an intrinsic mechanism for evaluating a tokenizer: for NLP tasks, the tokenizer which leads to the highest Rényi efficiency of the unigram distribution should be chosen. The Rényi efficiency is thus treated as a predictor of downstream performance (e.g., predicting BLEU for a machine translation task), without the expensive step of training multiple models with different tokenizers. Although useful, the predictive power of this metric is not perfect, and the authors note there are additional qualities of a good tokenization scheme that Rényi efficiency alone cannot capture. We describe two variants of BPE tokenization which can arbitrarily increase Rényi efficiency while decreasing the downstream model performance. These counterexamples expose cases where Rényi efficiency fails as an intrinsic tokenization metric and thus give insight for building more accurate predictors.

Entity linking is a well-established task in NLP consisting of associating entity mentions with entries in a knowledge base. Current models have demonstrated competitive performance in standard text settings. However, when it comes to noisy domains such as social media, certain challenges still persist. Typically, to evaluate entity linking on existing benchmarks, a comprehensive knowledge base is necessary and models are expected to possess an understanding of all the entities contained within the knowledge base. However, in practical scenarios where the objective is to retrieve sentences specifically related to a particular entity, strict adherence to a complete understanding of all entities in the knowledge base may not be necessary. To address this gap, we introduce TweetTER (Tweet Target Entity Retrieval), a novel benchmark that aims to bridge the challenges in entity linking. The distinguishing feature of this benchmark is its approach of re-framing entity linking as a binary entity retrieval task. This enables the evaluation of language models’ performance without relying on a conventional knowledge base, providing a more practical and versatile evaluation framework for assessing the effectiveness of language models in entity retrieval tasks.

Being labeled as a low-resource language, the Tunisian dialect has no existing prior TTS research. In this paper, we present a speech corpus for Tunisian Arabic Text-to-Speech (TunArTTS) to initiate the development of end-to-end TTS systems for the Tunisian dialect. Our Speech corpus is extracted from an online English and Tunisian Arabic dictionary. We were able to extract a mono-speaker speech corpus of +3 hours of a male speaker sampled at 44100 kHz. The corpus is processed and manually diacritized. Furthermore, we develop various TTS systems based on two approaches: training from scratch and transfer learning. Both Tacotron2 and FastSpeech2 were used and evaluated using subjective and objective metrics. The experimental results show that our best results are obtained with the transfer learning from a pre-trained model on the English LJSpeech dataset. This model obtained a mean opinion score (MOS) of 3.88. TunArTTS will be publicly available for research purposes along with the baseline TTS system demo. Keywords: Tunisian Dialect, Text-To-Speech, Low-resource, Transfer Learning, TunArTTS

Retrieval-augmented language models (RALMs) have demonstrated significant potential in refining and expanding their internal memory by retrieving evidence from external sources. However, RALMs will inevitably encounter knowledge conflicts when integrating their internal memory with external sources. Knowledge conflicts can ensnare RALMs in a tug-of-war between knowledge, limiting their practical applicability. In this paper, we focus on exploring and resolving knowledge conflicts in RALMs. First, we present an evaluation framework for assessing knowledge conflicts across various dimensions. Then, we investigate the behavior and preference of RALMs from the following two perspectives: (1) Conflicts between internal memory and external sources: We find that stronger RALMs emerge with the Dunning-Kruger effect, persistently favoring their faulty internal memory even when correct evidence is provided. Besides, RALMs exhibit an availability bias towards common knowledge; (2) Conflicts between truthful, irrelevant and misleading evidence: We reveal that RALMs follow the principle of majority rule, leaning towards placing trust in evidence that appears more frequently. Moreover, we find that RALMs exhibit confirmation bias, and are more willing to choose evidence that is consistent with their internal memory. To solve the challenge of knowledge conflicts, we propose a method called Conflict-Disentangle Contrastive Decoding (CD2) to better calibrate the model’s confidence. Experimental results demonstrate that our CD2 can effectively resolve knowledge conflicts in RALMs.

As Large Language Models (LLMs) become increasingly influential in reasoning tasks, ensuring their trustworthiness and introspective self-awareness is critical. This research introduces the Think-Solve-Verify (TSV) framework, an innovative strategy tailored to explore LLMs’ trustworthiness, introspective self-awareness, and collaborative reasoning. This method accentuates a model’s capability to construct introspective reasoning processes from answers and ensure their trustworthiness. The reasoning with TSV consistently performs at or near the top across the majority of datasets with a single interaction with LLM. Moreover, we refine the voting process of self-consistency within the Chain-of-Thought (CoT) approach, leading to notable accuracy enhancements. In our evaluations, this approach improved performance from 67.3% to 72.8% on the AQuA dataset. Furthermore, we delve into the model’s ability to explain the given answers, highlighting the significance of discerning genuine comprehension from mere guesswork.

Most data-to-text datasets are for English, so the difficulties of modelling data-to-text for low-resource languages are largely unexplored. In this paper we tackle data-to-text for isiXhosa, which is low-resource and agglutinative. We introduce Triples-to-isiXhosa (T2X), a new dataset based on a subset of WebNLG, which presents a new linguistic context that shifts modelling demands to subword-driven techniques. We also develop an evaluation framework for T2X that measures how accurately generated text describes the data. This enables future users of T2X to go beyond surface-level metrics in evaluation. On the modelling side we explore two classes of methods - dedicated data-to-text models trained from scratch and pretrained language models (PLMs). We propose a new dedicated architecture aimed at agglutinative data-to-text, the Subword Segmental Pointer Generator (SSPG). It jointly learns to segment words and copy entities, and outperforms existing dedicated models for 2 agglutinative languages (isiXhosa and Finnish). We investigate pretrained solutions for T2X, which reveals that standard PLMs come up short. Fine-tuning machine translation models emerges as the best method overall. These findings underscore the distinct challenge presented by T2X: neither well-established data-to-text architectures nor customary pretrained methodologies prove optimal. We conclude with a qualitative analysis of generation errors and an ablation study.