COLING 2024 论文列表

Making information accessible to diverse target audiences, including individuals with dyslexia and cognitive disabilities, is crucial. Automatic Text Simplification (ATS) systems aim to facilitate readability and comprehension by reducing linguistic complexity. However, they often lack customizability to specific user needs, and training data for smaller languages can be scarce. This paper addresses ATS in a Swedish context, using methods that provide more control over the simplification. A dataset of Swedish paraphrases is mined from large amounts of text and used to train ATS models utilizing prefix-tuning with control prefixes. We also introduce a novel data-driven method for selecting complexity attributes for controlling the simplification and compare it with previous approaches. Evaluation of the trained models using SARI and BLEU demonstrates significant improvements over the baseline — a fine-tuned Swedish BART model — and compared to previous Swedish ATS results. These findings highlight the effectiveness of employing paraphrase data in conjunction with controllable generation mechanisms for simplification. Additionally, the set of explored attributes yields similar results compared to previously used attributes, indicating their ability to capture important simplification aspects.

We developed a controllable paraphrase generation model for semantic and lexical similarities using a simple and intuitive mechanism: attaching tags to specify these values at the head of the input sentence. Lexically diverse paraphrases have been long coveted for data augmentation. However, their generation is not straightforward because diversifying surfaces easily degrades semantic similarity. Furthermore, our experiments revealed two critical features in data augmentation by paraphrasing: appropriate similarities of paraphrases are highly downstream task-dependent, and mixing paraphrases of various similarities negatively affects the downstream tasks. These features indicated that the controllability in paraphrase generation is crucial for successful data augmentation. We tackled these challenges by fine-tuning a pre-trained sequence-to-sequence model employing tags that indicate the semantic and lexical similarities of synthetic paraphrases selected carefully based on the similarities. The resultant model could paraphrase an input sentence according to the tags specified. Extensive experiments on data augmentation for contrastive learning and pre-fine-tuning of pretrained masked language models confirmed the effectiveness of the proposed model. We release our paraphrase generation model and a corpus of 87 million diverse paraphrases. (https://github.com/Ogamon958/ConPGS)

The present work studies the contribution of move structure to automatic genre identification. This concept - well known in other branches of genre analysis - seems to have little application in natural language processing. We describe how we collect a corpus of websites in French related to tourism and annotate it with move structure. We conduct experiments on automatic genre identification with our corpus. Our results show that our approach for informing a model with move structure can increase its performance for automatic genre identification, and reduce the need for annotated data and computational power.

This paper presents ContrastWSD, a RoBERTa-based metaphor detection model that integrates the Metaphor Identification Procedure (MIP) and Word Sense Disambiguation (WSD) to extract and contrast the contextual meaning with the basic meaning of a word to determine whether it is used metaphorically in a sentence. By utilizing the word senses derived from a WSD model, our model enhances the metaphor detection process and outperforms other methods that rely solely on contextual embeddings or integrate only the basic definitions and other external knowledge. We evaluate our approach on various benchmark datasets and compare it with strong baselines, indicating the effectiveness in advancing metaphor detection.

Topic model is a statistical model that leverages unsupervised learning to mine hidden topics in document collections. The data sparsity and colloquialism of social texts make it difficult to accurately mine the topics. Traditional methods assume that there are only 0/1-state relationships between the two parties in the social networks, but the relationship status in real life is more complicated, such as continuously changing relationships with different degrees of intimacy. This paper proposes a continuous relational diffusion driven topic model (CRTM) with multi-grained text for microblog to realize the continuous representation of the relationship state and make up for the context and structural information lost by previous representation methods. Multi-grained text representation learning distinguishes the impact of formal and informal expression on the topics further and alleviates colloquialism problems. Specifically, based on the original social network, the reconstructed social network with continuous relationship status is obtained by using information diffusion technology. The graph convolution model is utilized to learn node embeddings through the new social network. Finally, the neural variational inference is applied to generate topics according to continuous relationships. We validate CRTM on three real datasets, and the experimental results show the effectiveness of the scheme.

This paper introduces Continued Pre-training on Analogies (CPoA) to incorporate pre-trained language models with analogical abilities, aiming at improving performance in low-resource translations without data augmentation. We continue training the models on sentence analogies retrieved from a translation corpus. Considering the sparsity of analogy in corpora, especially in low-resource scenarios, we propose exploring approximate analogies between sentences. We attempt to find sentence analogies that might not conform to formal criteria for entire sentences but partial pieces. When training the models, we introduce a weighting scalar pertaining to the quality of analogies to adjust the influence: emphasizing closer analogies while diminishing the impact of far ones. We evaluate our approach on a low-resource translation task: German-Upper Sorbian. The results show that CPoA using 10 times fewer instances can effectively attain gains of +1.4 and +1.3 BLEU points over the original model in two translation directions. This improvement is more pronounced when there are fewer parallel examples.

Controlled Text Generation (CTG) steers the generation of continuations of a given context (prompt) by a Large Language Model (LLM) towards texts possessing a given attribute (e.g., topic, sentiment). In this paper we view CTG as a Continual Learning problem: how to learn at every step to steer next-word generation, without having to wait for end-of-sentence. This continual view is useful for online applications such as CTG for speech, where end-of-sentence is often uncertain. We depart from an existing model, the Plug-and-Play language models (PPLM), which perturbs the context at each step to better predict next-words that posses the desired attribute. While PPLM is intricate and has many hyper-parameters, we provide a proof that the PPLM objective function can be reduced to a Continual Reinforcement Learning (CRL) reward function, thereby simplifying PPLM and endowing it with a better understood learning framework. Subsequently, we present, the first of its kind, CTG algorithm that is fully based on CRL and exhibit promising empirical results.

Traditional continual event detection relies on abundant labeled data for training, which is often impractical to obtain in real-world applications. In this paper, we introduce continual few-shot event detection (CFED), a more commonly encountered scenario when a substantial number of labeled samples are not accessible. The CFED task is challenging as it involves memorizing previous event types and learning new event types with few-shot samples. To mitigate these challenges, we propose a memory-based framework: Hierarchical Augmentation Network (HANet). To memorize previous event types with limited memory, we incorporate prototypical augmentation into the memory set. For the issue of learning new event types in few-shot scenarios, we propose a contrastive augmentation module for token representations. Despite comparing with previous state-of-the-art methods, we also conduct comparisons with ChatGPT. Experiment results demonstrate that our method significantly outperforms all of these methods in multiple continual few-shot event detection tasks.

Abstractive citation text generation is usually framed as an infilling task, where a sequence-to-sequence model is trained to generate a citation given a reference paper and the context window around the target; the generated citation should be a brief discussion of the reference paper as it relates to the citing context. However, examining a recent LED-based citation generation system, we find that many of the generated citations are generic summaries of the reference paper’s main contribution, ignoring the citation context’s focus on a different topic. To address this problem, we propose a simple modification to the citation text generation task: the generation target is not only the citation itself, but the entire context window, including the target citation. This approach can be easily applied to any abstractive citation generation system, and our experimental results show that training in this way is preferred by human readers and allows the generation model to make use of contextual clues about what topic to discuss and what stance to take.

Contextual information, including the sentences in the same document and in other documents of the dataset, plays a crucial role in improving the accuracy of document-level ASR Error Correction (AEC), while most previous works ignore this. In this paper, we propose a context-aware method that utilizes a k-Nearest Neighbors (kNN) approach to enhance the AEC model by retrieving a datastore containing contextual information. We conduct experiments on two English and two Chinese datasets, and the results demonstrate that our proposed model can effectively utilize contextual information to improve document-level AEC. Furthermore, the context information from the whole dataset provides even better results.

Previous studies employ the autoregressive translation (AT) paradigm in the document-to-document neural machine translation. These methods extend the translation unit from a single sentence to a pseudo-document and encodes the full pseudo-document, avoiding the redundant computation problem in context. However, the AT methods cannot parallelize decoding and struggle with error accumulation, especially when the length of sentences increases. In this work, we propose a context-aware non-autoregressive framework with the sentence-aligned connectionist temporal classification (SA-CTC) loss for document-level neural machine translation. In particular, the SA-CTC loss reduces the search space of the decoding path by fixing the positions of the beginning and end tokens for each sentence in the document. Meanwhile, the context-aware architecture introduces preset nodes to represent sentence-level information and utilizes a hierarchical attention structure to regulate the attention hypothesis space. Experimental results show that our proposed method can achieve competitive performance compared with several strong baselines. Our method implements non-autoregressive modeling in Doc-to-Doc translation manner, achieving an average 46X decoding speedup compared to the document-level AT baselines on three benchmarks.

We study the communication of concepts at different levels of abstraction and in different contexts in an agent-based, interactive reference game. While playing the concept-level reference game, the neural network agents develop a communication system from scratch. We use a novel symbolic dataset that disentangles concept type (ranging from specific to generic) and context (ranging from fine to coarse) to study the influence of these factors on the emerging language. We compare two game scenarios: one in which speaker agents have access to context information (context-aware) and one in which the speaker agents do not have access to context information (context-unaware). First, we find that the agents learn higher-level concepts from the object inputs alone. Second, an analysis of the emergent communication system shows that only context-aware agents learn to communicate efficiently by adapting their messages to the context conditions and relying on context for unambiguous reference. Crucially, this behavior is not explicitly incentivized by the game, but efficient communication emerges and is driven by the availability of context alone. The emerging language we observe is reminiscent of evolutionary pressures on human languages and highlights the pivotal role of context in a communication system.

Despite the remarkable achievements of Large Language Models (LLMs) in various Natural Language Processing tasks, their competence in abstract language understanding remains a relatively under-explored territory. Figurative language interpretation serves as ideal testbed for assessing this as it requires models to navigate beyond the literal meaning and delve into underlying semantics of the figurative expressions. In this paper, we seek to examine the performance of GPT-3.5 in zero-shot setting through word-level metaphor detection. Specifically, we frame the task as annotation of word-level metaphors in proverbs. To this end, we employ a dataset of English proverbs and evaluated its performance by applying different prompting strategies. Our results show that the model shows a satisfactory performance at identifying word-level metaphors, particularly when it is prompted with a hypothetical context preceding the proverb. This observation underscores the pivotal role of well-designed prompts for zero-shot settings through which these models can be leveraged as annotators for subjective NLP tasks.

In this paper, we make a contribution that can be understood from two perspectives: from an NLP perspective, we introduce a small challenge dataset for NLI with large lexical overlap, which minimises the possibility of models discerning entailment solely based on token distinctions, and show that GPT-4 and Llama 2 fail it with strong bias. We then create further challenging sub-tasks in an effort to explain this failure. From a Computational Linguistics perspective, we identify a group of constructions with three classes of adjectives which cannot be distinguished by surface features. This enables us to probe for LLM’s understanding of these constructions in various ways, and we find that they fail in a variety of ways to distinguish between them, suggesting that they don’t adequately represent their meaning or capture the lexical properties of phrasal heads.

Datasets that pair Knowledge Graphs (KG) and text together (KG-T) can be used to train forward and reverse neural models that generate text from KG and vice versa. However models trained on datasets where KG and text pairs are not equivalent can suffer from more hallucination and poorer recall. In this paper, we verify this empirically by generating datasets with different levels of noise and find that noisier datasets do indeed lead to more hallucination. We argue that the ability of forward and reverse models trained on a dataset to cyclically regenerate source KG or text is a proxy for the equivalence between the KG and the text in the dataset. Using cyclic evaluation we find that manually created WebNLG is much better than automatically created TeKGen and T-REx. Informed by these observations, we construct a new, improved dataset called LAGRANGE using heuristics meant to improve equivalence between KG and text and show the impact of each of the heuristics on cyclic evaluation. We also construct two synthetic datasets using large language models (LLMs), and observe that these are conducive to models that perform significantly well on cyclic generation of text, but less so on cyclic generation of KGs, probably because of a lack of a consistent underlying ontology.

We introduce a manually annotated syntactic treebank based on Universal Dependencies, derived from the written data of second language (L2) Korean learners. In developing this new dataset, we critically evaluated previous works and revised the annotation guidelines to better reflect the linguistic properties of Korean and the characteristics of L2 learners. The L2 Korean treebank encompasses 7,530 sentences (66,982 words; 129,333 morphemes) and is publicly available at: https://github.com/NLPxL2Korean/L2KW-corpus.

Multilingual L2 speech corpora for developing automatic speech assessment are currently available, but they lack comprehensive annotations of L2 speech from non-native speakers of various languages. This study introduces the methodology of designing a Korean learners’ L2 speech corpus of seven languages: English, Japanese, Chinese, French, German, Spanish, and Russian. We describe the development of reading scripts, reading tasks, scoring criteria, and expert evaluation methods in detail. Our corpus contains 1,200 hours of L2 speech data from Korean learners (400 hours for English, 200 hours each for Japanese and Chinese, 100 hours each for French, German, Spanish, and Russian). The corpus is annotated with spelling and pronunciation transcription, expert pronunciation assessment scores (accuracy of pronunciation and fluency of prosody), and metadata such as gender, age, self-reported language proficiency, and pronunciation error types. We also propose a practical verification method and a reliability threshold to ensure the reliability and objectivity of large-scale subjective evaluation data.

Research in low-resource language is often hampered due to the under-representation of how the language is being used in reality. This is particularly true for Indonesian language because there is a limited variety of textual datasets, and majority were acquired from official sources with formal writing style. All the more for the task of geoparsing, which could be implemented for navigation and travel planning applications, such datasets are rare, even in the high-resource languages, such as English. Being aware of the need for a new resource in both languages for this specific task, we constructed a new dataset comprising both Indonesian and English from personal travelogue articles. Our dataset consists of 88 articles, exactly half of them written in each language. We covered both named and nominal expressions of four entity types related to travel: location, facility, transportation, and line. We also conducted experiments by training classifiers to recognise named entities and their nominal expressions. The results of our experiments showed a promising future use of our dataset as we obtained F1-score above 0.9 for both languages.

Contrary to common belief, there are rich and diverse data sources available for many thousands of languages, which can be used to develop technologies for these languages. In this paper, we provide an overview of some of the major online data sources, the types of data that they provide access to, potential applications of this data, and the number of languages that they cover. Even this covers only a small fraction of the data that exists; for example, printed books are published in many languages but few online aggregators exist.