COLING 2024 论文列表

This paper presents our work on a task of automatic decomposition of text editing examples into primitive edit operations. Toward a detailed analysis of the behavior of text editing systems, identification of fine-grained edit operations performed by the systems is essential. Given a pair of source and edited sentences, the goal of our task is to generate a non-redundant sequence of primitive edit operations, i.e., the semantically minimal edit operations preserving grammaticality, that iteratively converts the source sentence to the edited sentence. First, we formalize this task, explaining its significant features and specifying the constraints that primitive edit operations should satisfy. Then, we propose a method to automate this task, which consists of two steps: generation of an edit operation lattice and selection of an optimal path. To obtain a wide range of edit operation candidates in the first step, we combine a phrase aligner and a large language model. Experimental results show that our method perfectly decomposes 44% and 64% of editing examples in the text simplification and machine translation post-editing datasets, respectively. Detailed analyses also provide insights into the difficulties of this task, suggesting directions for improvement.

Data visualization has emerged as an effective tool for getting insights from massive datasets. Due to the hardness of manipulating the programming languages of data visualization, automatic data visualization generation from natural languages (Text-to-Vis) is becoming increasingly popular. Despite the plethora of research effort on the English Text-to-Vis, studies have yet to be conducted on data visualization generation from questions in Chinese. Motivated by this, we propose a Chinese Text-to-Vis dataset in the paper and demonstrate our first attempt to tackle this problem. Our model integrates multilingual BERT as the encoder, boosts the cross-lingual ability, and infuses the n-gram information into our word representation learning. Our experimental results show that our dataset is challenging and deserves further research.

Geoparsing is the task of estimating the latitude and longitude (coordinates) of location expressions in texts. Geoparsing must deal with the ambiguity of the expressions that indicate multiple locations with the same notation. For evaluating geoparsing systems, several corpora have been proposed in previous work. However, these corpora are small-scale and suffer from the coverage of location expressions on general domains. In this paper, we propose Wikipedia Hyperlink-based Location Linking (WHLL), a novel method to construct a large-scale corpus for geoparsing from Wikipedia articles. WHLL leverages hyperlinks in Wikipedia to annotate multiple location expressions with coordinates. With this method, we constructed the WHLL corpus, a new large-scale corpus for geoparsing. The WHLL corpus consists of 1.3M articles, each containing about 7.8 unique location expressions. 45.6% of location expressions are ambiguous and refer to more than one location with the same notation. In each article, location expressions of the article title and those hyperlinks to other articles are assigned with coordinates. By utilizing hyperlinks, we can accurately assign location expressions with coordinates even with ambiguous location expressions in the texts. Experimental results show that there remains room for improvement by disambiguating location expressions.

Aspect Sentiment Triplet Extraction (ASTE), introduced in 2020, is a task that involves the extraction of three key elements: target aspects, descriptive opinion spans, and their corresponding sentiment polarity. This process, however, faces a significant hurdle, particularly when applied to Chinese languages, due to the lack of sufficient datasets for model training, largely attributable to the arduous manual labeling process. To address this issue, we present an innovative framework that facilitates the automatic construction of ASTE via Iterative Weak Supervision, negating the need for manual labeling, aided by a discriminator to weed out subpar samples. The objective is to successively improve the quality of this raw data and generate supplementary data. The effectiveness of our approach is underscored by our results, which include the creation of a substantial Chinese review dataset. This dataset encompasses over 60,000 Google restaurant reviews in Chinese and features more than 200,000 extracted triplets. Moreover, we have also established a robust baseline model by leveraging a novel method of weak supervision. Both our dataset and model are openly accessible to the public.

One of the most important communicative skills children have to learn is to engage in meaningful conversations with people around them. At the heart of this learning lies the mastery of contingency, i.e., the ability to contribute to an ongoing exchange in a relevant fashion (e.g., by staying on topic). Current research on this question relies on the manual annotation of a small sample of children, which limits our ability to draw general conclusions about development. Here, we propose to mitigate the limitations of manual labor by relying on automatic tools for contingency judgment in children’s early natural interactions with caregivers. Drawing inspiration from the field of dialogue systems evaluation, we built and compared several automatic classifiers. We found that a Transformer-based pre-trained language model – when fine-tuned on a relatively small set of data we annotated manually (around 3,500 turns) – provided the best predictions. We used this model to automatically annotate, new and large-scale data, almost two orders of magnitude larger than our fine-tuning set. It was able to replicate existing results and generate new data-driven hypotheses. The broad impact of the work is to provide resources that can help the language development community study communicative development at scale, leading to more robust theories.

As the quality of AI-generated text increases with the development of new Large Language Models, people use them to write in a variety of contexts. Human-AI collaborative writing poses a potential challenge for existing AI analysis techniques, which have been primarily tested either on human-written text only, or on samples independently generated by humans and AI. In this work, we investigate the extent to which existing AI detection and authorship analysis models can perform classification on data generated in human-AI collaborative writing sessions. Results show that, for AI text detection in the cowriting setting, classifiers based on authorship embeddings (Rivera-Soto et al., 2021) outperform classifiers used in prior work distinguishing AI vs. human text generated independently. However, these embeddings are not optimal for finer-grained authorship identification tasks: for authorship verification, n-gram based models are more robust to human-AI co-written text, and authorship attribution performance degrades compared to baselines that use human-written text only. Taken together, this suggests that the rise of human-AI co-written text will require adapting AI detection tools and authorship analysis techniques in the near future. We release our code at https://github.com/AARichburg/Human-AI_Authorship_Analysis.

The acquisition of grammar has been a central question to adjudicate between theories of language acquisition. In order to conduct faster, more reproducible, and larger-scale corpus studies on grammaticality in child-caregiver conversations, tools for automatic annotation can offer an effective alternative to tedious manual annotation. We propose a coding scheme for context-dependent grammaticality in child-caregiver conversations and annotate more than 4,000 utterances from a large corpus of transcribed conversations. Based on these annotations, we train and evaluate a range of NLP models. Our results show that fine-tuned Transformer-based models perform best, achieving human inter-annotation agreement levels. As a first application and sanity check of this tool, we use the trained models to annotate a corpus almost two orders of magnitude larger than the manually annotated data and verify that children’s grammaticality shows a steady increase with age. This work contributes to the growing literature on applying state-of-the-art NLP methods to help study child language acquisition at scale.

We introduce the first Romanian animacy classifier, specifically a type-based binary classifier of Romanian nouns into the classes human/non-human, using pre-trained word embeddings and animacy information derived from Romanian WordNet. By obtaining a seed set of labeled nouns and their embeddings, we are able to train classifiers that generalize to unseen nouns. We compare three different architectures and observe good performance on classifying word types. In addition, we manually annotate a small corpus for animacy to perform a token-based evaluation of Romanian animacy classification in a naturalistic setting, which reveals limitations of the type-based classification approach.

As in oral phonology, prosody is an important carrier of linguistic information in sign languages. One of the most prominent ways this reveals itself is in the time structure of signs: their rhythm and intensity of articulation. To be able to empirically see these effects, the velocity of the hands can be computed throughout the execution of a sign. In this article, we propose a method for extracting this information from unlabeled videos of sign language, exploiting CoTracker, a recent advancement in computer vision which can track every point in a video without the need of any calibration or fine-tuning. The dominant hand is identified via clustering of the computed point velocities, and its dynamic profile plotted to make apparent the prosodic structure of signing. We apply our method to different datasets and sign languages, and perform a preliminary visual exploration of results. This exploration supports the usefulness of our methodology for linguistic analysis, though issues to be tackled remain, such as bi-manual signs and a formal and numerical evaluation of accuracy. Nonetheless, the absence of any preprocessing requirements may make it useful for other researchers and datasets.

The advent of self-supervised learning (SSL) in speech processing has allowed the use of large unlabeled datasets to learn pre-trained models, serving as powerful encoders for various downstream tasks. However, the application of these SSL methods to languages such as French has proved difficult due to the scarcity of large French speech datasets. To advance the emergence of pre-trained models for French speech, we present the Audiocite.net corpus composed of 6,682 hours of recordings from 130 readers. This corpus is composed of audiobooks from the audiocite.net website, shared by 130 readers. In addition to describing the creation process and final statistics, we also show how this dataset impacted the models of LeBenchmark project in its 14k version for speech processing downstream tasks.

Large language models (LLMs) empowered by chain-of-thought (CoT) prompting have yielded remarkable prowess in reasoning tasks. Nevertheless, current methods predominantly lean on handcrafted or task-specific demonstrations, lack reliable knowledge basis and thus struggle for trustworthy responses in an automated pattern. While recent works endeavor to improve upon one certain aspect, they ignore the importance and necessity of establishing an integrated and interpretable reasoning system. To address these drawbacks and provide a universal solution, we propose AuRoRA: a one-for-all platform for augmented reasoning and refining based on CoT prompting that excels in adaptability, reliability, integrity, and interpretability. The system exhibits superior performances across six reasoning tasks and offers real-time visual analysis, which has pivotal academic and application value in the era of LLMs. The AuRoRA platform is available at https://huggingface.co/spaces/Anni123/AuRoRA.

Named Entity Recognition (NER) is a cornerstone natural language processing task while its robustness has been given little attention. This paper rethinks the principles of the conventional text attack, as they can easily violate the label consistency between the original and adversarial NER samples. This is due to the fine-grained nature of NER, as even minor word changes in the sentence can result in the emergence or mutation of any entity, producing invalid adversarial samples. To this end, we propose a novel one-word modification NER attack based on a key insight, NER models are always vulnerable to the boundary position of an entity to make their decision. We thus strategically insert a new boundary into the sentence and trigger the victim model to make a wrong recognition either on this boundary word or on other words in the sentence. We call this attack Virtual Boundary Attack (ViBA), which is shown to be remarkably effective when attacking both English and Chinese models with a 70%-90% attack success rate on state-of-the-art language models, and also significantly faster than previous methods.

The study of ancient Middle Eastern cultures is dominated by the vast number of cuneiform texts. Multiple languages and language families were expressed in cuneiform. The most dominant language written in cuneiform is the Semitic Akkadian, which is the focus of this paper. We are specifically focusing on letters written in the dialect used in modern-day Baghdad and south towards the Persian Gulf during the Old Babylonian period (c. 2000-1600 B.C.E.). The Akkadian language was rediscovered in the 19th century and is now being scrutinised by Natural Language Processing (NLP) methods. However, existing Akkadian text publications are not always suitable for digital editions. We therefore risk applying NLP methods onto renderings of Akkadian unfit for the purpose. In this paper we want to investigate the input material and try to initiate a discussion about best-practices in the crossroad where NLP meets cuneiform studies. Specifically, we want to question the use of pre-trained embeddings, sentence segmentation and the type of cuneiform input used to fine-tune language models for the task of fine-grained part-of-speech tagging. We examine the issues by theoretical and practical approaches in a way that we hope spurs discussions that are relevant for automatic processing of other ancient languages.

In this study, we evaluated the performance of the state-of-the-art sequence tagging grammar error detection and correction model (SeqTagger) using Japanese university students’ writing samples. With an automatic annotation toolkit, ERRANT, we first evaluated SeqTagger’s performance on error correction with human expert correction as the benchmark. Then a human-annotated approach was adopted to evaluate Seqtagger’s performance in error detection using a subset of the writing dataset. Results indicated a precision of 63.66% and a recall of 20.19% for error correction in the full dataset. For the subset, after manual exclusion of irrelevant errors such as semantic and mechanical ones, the model shows an adjusted precision of 97.98% and an adjusted recall of 42.98% for error detection, indicating the model’s high accuracy but also its conservativeness. Thematic analysis on errors undetected by the model revealed that determiners and articles, especially the latter, were predominant. Specifically, in terms of context-independent errors, the model occasionally overlooked basic ones and faced challenges with overly erroneous or complex structures. Meanwhile, context-dependent errors, notably those related to tense and noun number, as well as those possibly influenced by the students’ first language (L1), remained particularly challenging.

This paper offers a nuanced examination of the role Large Language Models (LLMs) play in coreference resolution, aimed at guiding the future direction in the era of LLMs. We carried out both manual and automatic analyses of different LLMs’ abilities, employing different prompts to examine the performance of different LLMs, obtaining a comprehensive view of their strengths and weaknesses. We found that LLMs show exceptional ability in understanding coreference. However, harnessing this ability to achieve state of the art results on traditional datasets and benchmarks isn’t straightforward. Given these findings, we propose that future efforts should: (1) Improve the scope, data, and evaluation methods of traditional coreference research to adapt to the development of LLMs. (2) Enhance the fine-grained language understanding capabilities of LLMs.

Providing constructive feedback on student essays is a critical factor in improving educational results; however, it presents notable difficulties and may demand substantial time investments, especially when aiming to deliver individualized and informative guidance. This study undertakes a comparative analysis of two readily available online resources for students seeking to hone their skills in essay writing for English proficiency tests: 1) essayforum.com, a widely used platform where students can submit their essays and receive feedback from volunteer educators at no cost, and 2) Large Language Models (LLMs) such as ChatGPT. By contrasting the feedback obtained from these two resources, we posit that they can mutually reinforce each other and are more helpful if employed in conjunction when seeking no-cost online assistance. The findings of this research shed light on the challenges of providing personalized feedback and highlight the potential of AI in advancing the field of automated essay evaluation.

This paper presents a novel back transliteration dataset capturing native language text originally composed in the Roman/Latin script, harvested from popular social media platforms, along with its corresponding representation in the native Assamese script. Assamese, categorized as a low-resource language within the Indo-Aryan language family, predominantly spoken in the north-east Indian state of Assam, faces a scarcity of linguistic resources. The dataset comprises a total of 60,312 Roman-native parallel transliterated sentences. This paper diverges from conventional forward transliteration datasets consisting mainly of named entities and technical terms, instead presenting a novel transliteration dataset cultivated from three prominent social media platforms, Facebook, Twitter(currently X), and YouTube, in the backward transliteration direction. The paper offers a comprehensive examination of ten state-of-the-art word-level transliteration models within the context of this dataset, encompassing transliteration evaluation benchmarks, extensive performance assessments, and a discussion of the unique challenges encountered during the processing of transliterated social media content. Our approach involves the initial use of two statistical transliteration models, followed by the training of two state-of-the-art neural network-based transliteration models, evaluation of three publicly available pre-trained models, and ultimately fine-tuning one existing state-of-the-art multilingual transliteration model along with two pre-trained large language models using the collected datasets. Notably, the Neural Transformer model outperforms all other baseline transliteration models, achieving the lowest Word Error Rate (WER) and Character Error Rate (CER), and the highest BLEU (up to 4 gram) score of 55.05, 19.44, and 69.15, respectively.

This paper presents a question-answering approach to extract document-level event-argument structures. We automatically ask and answer questions for each argument type an event may have. Questions are generated using manually defined templates and generative transformers. Template-based questions are generated using predefined role-specific wh-words and event triggers from the context document. Transformer-based questions are generated using large language models trained to formulate questions based on a passage and the expected answer. Additionally, we develop novel data augmentation strategies specialized in inter-sentential event-argument relations. We use a simple span-swapping technique, coreference resolution, and large language models to augment the training instances. Our approach enables transfer learning without any corpora-specific modifications and yields competitive results with the RAMS dataset. It outperforms previous work, and it is especially beneficial to extract arguments that appear in different sentences than the event trigger. We also present detailed quantitative and qualitative analyses shedding light on the most common errors made by our best model.

Classifying research output into context-specific label taxonomies is a challenging and relevant downstream task, given the volume of existing and newly published articles. We propose a method to enhance the performance of article classification by enriching simple Graph Neural Network (GNN) pipelines with multi-graph representations that simultaneously encode multiple signals of article relatedness, e.g. references, co-authorship, shared publication source, shared subject headings, as distinct edge types. Fully supervised transductive node classification experiments are conducted on the Open Graph Benchmark OGBN-arXiv dataset and the PubMed diabetes dataset, augmented with additional metadata from Microsoft Academic Graph and PubMed Central, respectively. The results demonstrate that multi-graphs consistently improve the performance of a variety of GNN models compared to the default graphs. When deployed with SOTA textual node embedding methods, the transformed multi-graphs enable simple and shallow 2-layer GNN pipelines to achieve results on par with more complex architectures.