COLING 2024 Papers

One of the most significant pieces of ancient Greek literature, the Iliad, is part of humanity’s collective cultural heritage. This work aims to provide the scientific community with an emotion-labeled dataset for classical literature and Western mythology in particular. To model the emotions of the poem, we use a multi-variate time series. We also evaluated the dataset by means of two methods. We compare the manual classification against a dictionary-based benchmark as well as employ a state-of-the-art deep learning masked language model that has been tuned using our data. Both evaluations return encouraging results (MSE and MAE Macro Avg 0.101 and 0.188 respectively) and highlight some interesting phenomena.

Bias in NLP models may arise from using pre-trained transformer models trained on biased corpora, or by training or fine-tuning directly on corpora with systemic biases. Recent research has explored strategies for reduce measurable biases in NLP predictions while maintaining prediction accuracy on held-out test sets, e.g., by modifying word embedding geometry after training, using purpose-built neural modules for training, or automatically augmenting training data with examples designed to reduce bias. This paper focuses on a debiasing strategy for aspect-based sentiment analysis (ABSA) by augmenting the training data using norm-based language templates derived from previous language resources. We show that the baseline model predicts lower sentiment toward some topics and individuals than others and has relatively high prediction bias (measured by standard deviation), even when the context is held constant. Our results show that our norm-based data augmentation reduces topical bias to less than half while maintaining prediction quality (measured by RMSE), by augmenting the training data by only 1.8%.

This paper focuses on dealing with data scarcity in spoken question answering (QA) using automatic question-answer generation and a carefully selected fine-tuning strategy that leverages limited annotated data (paragraphs and question-answer pairs). Spoken QA is a challenging task due to using spoken documents, i.e., erroneous automatic speech recognition (ASR) transcriptions, and the scarcity of spoken QA data. We propose a framework for utilizing limited annotated data effectively to improve spoken QA performance. To deal with data scarcity, we train a question-answer generation model with annotated data and then produce large amounts of question-answer pairs from unannotated data (paragraphs). Our experiments demonstrate that incorporating limited annotated data and the automatically generated data through a carefully selected fine-tuning strategy leads to 5.5% relative F1 gain over the model trained only with annotated data. Moreover, the proposed framework is also effective in high ASR errors.

In today’s rapidly evolving digital age, disinformation poses a significant threat to public sentiment and socio-political dynamics. To address this, we introduce a new dataset “DeFaktS”, designed to understand and counter disinformation within German media. Distinctively curated across various news topics, DeFaktS offers an unparalleled insight into the diverse facets of disinformation. Our dataset, containing 105,855 posts with 20,008 meticulously labeled tweets, serves as a rich platform for in-depth exploration of disinformation’s diverse characteristics. A key attribute that sets DeFaktS apart is, its fine-grain annotations based on polarized categories. Our annotation framework, grounded in the textual characteristics of news content, eliminates the need for external knowledge sources. Unlike most existing corpora that typically assign a singular global veracity value to news, our methodology seeks to annotate every structural component and semantic element of a news piece, ensuring a comprehensive and detailed understanding. In our experiments, we employed a mix of classical machine learning and advanced transformer-based models. The results underscored the potential of DeFaktS, with transformer models, especially the German variant of BERT, exhibiting pronounced effectiveness in both binary and fine-grained classifications.

Extracting who says what to whom is a crucial part in analyzing human communication in today’s abundance of data such as online news articles. Yet, the lack of annotated data for this task in German news articles severely limits the quality and usability of possible systems. To remedy this, we present a new, freely available, creative-commons-licensed dataset for quotation attribution in German news articles based on WIKINEWS. The dataset provides curated, high-quality annotations across 1000 documents (250,000 tokens) in a fine-grained annotation schema enabling various downstream uses for the dataset. The annotations not only specify who said what but also how, in which context, to whom and define the type of quotation. We specify our annotation schema, describe the creation of the dataset and provide a quantitative analysis. Further, we describe suitable evaluation metrics, apply two existing systems for quotation attribution, discuss their results to evaluate the utility of our dataset and outline use cases of our dataset in downstream tasks.

Users of social media platforms are negatively affected by the proliferation of hate or abusive content. There has been a rise in homophobic and transphobic content in recent years targeting LGBT+ individuals. The increasing levels of homophobia and transphobia online can make online platforms harmful and threatening for LGBT+ persons, potentially inhibiting equality, diversity, and inclusion. We are introducing a new dataset for three languages, namely Telugu, Kannada, and Gujarati. Additionally, we have created an expert-labeled dataset to automatically identify homophobic and transphobic content within comments collected from YouTube. We provided comprehensive annotation rules to educate annotators in this process. We collected approximately 10,000 comments from YouTube for all three languages. Marking the first dataset of these languages for this task, we also developed a baseline model with pre-trained transformers.

Attention mechanisms play a crucial role in the neural revolution of Natural Language Processing (NLP). With the growth of attention-based models, several pruning techniques have been developed to identify and exploit sparseness, making these models more efficient. Most efforts focus on hard-coding attention patterns or pruning attention weights based on training data. We propose Attention Pruning (AP), a framework that observes attention patterns in a fixed dataset and generates a global sparseness mask. AP saves 90% of attention computation for language modeling and about 50% for machine translation and GLUE tasks, maintaining result quality. Our method reveals important distinctions between self- and cross-attention patterns, guiding future NLP research. Our framework can reduce both latency and memory requirements for any attention-based model, aiding in the development of improved models for existing or new NLP applications. We have demonstrated this with encoder and autoregressive transformer models using Triton GPU kernels and make our code publicly available at https://github.com/irugina/AP

Clinical NLP research faces a scarcity of publicly available datasets due to privacy concerns. MIMIC-III marked a significant milestone, enabling substantial progress, and now, with MIMIC-IV, the dataset has expanded significantly, offering a broader scope. In this paper, we focus on the task of predicting clinical outcomes from clinical text. This is crucial in modern healthcare, aiding in preventive care, differential diagnosis, and capacity planning. We introduce a novel clinical outcome prediction dataset derived from MIMIC-IV. Furthermore, we provide initial insights into the performance of models trained on MIMIC-III when applied to our new dataset, with specific attention to potential data drift. We investigate challenges tied to evolving documentation standards and changing codes in the International Classification of Diseases (ICD) taxonomy, such as the transition from ICD-9 to ICD-10. We also explore variations in clinical text across different hospital wards. Our study aims to probe the robustness and generalization of clinical outcome prediction models, contributing to the ongoing advancement of clinical NLP in healthcare.

This paper proposes Labadain Crawler, a data collection pipeline tailored to automate and optimize the process of constructing textual corpora from the web, with a specific target to low-resource languages. The system is built on top of Nutch, an open-source web crawler and data extraction framework, and incorporates language processing components such as a tokenizer and a language identification model. The pipeline efficacy is demonstrated through successful testing with Tetun, one of Timor-Leste’s official languages, resulting in the construction of a high-quality Tetun text corpus comprising 321.7k sentences extracted from over 22k web pages. The contributions of this paper include the development of a Tetun tokenizer, a Tetun language identification model, and a Tetun text corpus, marking an important milestone in Tetun text information retrieval.

In recent years, the dominance of Large Language Models (LLMs) in the English language has become evident. However, there remains a pronounced gap in resources and evaluation tools tailored for non-English languages, underscoring a significant disparity in the global AI landscape. This paper seeks to bridge this gap, specifically focusing on the Italian linguistic context. We introduce a novel benchmark, and an open LLM Leaderboard, designed to evaluate LLMs’ performance in Italian, providing a rigorous framework for comparative analysis. In our assessment of currently available models, we highlight their respective strengths and limitations against this standard. Crucially, we propose “DanteLLM”, a state-of-the-art LLM dedicated to Italian. Our empirical evaluations underscore Dante’s superiority, as it emerges as the most performant model on our benchmark, with improvements by up to 6 points. This research not only marks a significant stride in Italian-centric natural language processing but also offers a blueprint for the development and evaluation of LLMs in other languages, championing a more inclusive AI paradigm. Our code at: https://github.com/RSTLess-research/DanteLLM

The TableTextQA task requires finding the answer to the question from a combination of tabular and textual data, which has been gaining increasing attention. The row-based approaches have demonstrated remarkable effectiveness. However, they suffer from the following limitations: (1) a lack of interaction between rows; (2) excessively long input lengths; and (3) question attention shifts in the multi-hop QA task. To this end, we propose a novel method: Dynamic Multi-Granularity Graph Estimate Retrieval - DRAMA. Our method incorporates an interaction mechanism among multiple rows. Specifically, we utilize a memory bank to store the features of each row, thereby facilitating the construction of a heterogeneous graph with multi-row information. Besides, a Dynamic Graph Attention Network (DGAT) module is engaged to gauge the attention shift in the multi-hop question and eliminate the noise information dynamically. Empirical results on the widely used HybridQA and TabFact datasets demonstrate that the proposed model is effective.

Continuous Relation Extraction (CRE) aims to incrementally learn relation knowledge from a non-stationary stream of data. Since the introduction of new relational tasks can overshadow previously learned information, catastrophic forgetting becomes a significant challenge in this domain. Current replay-based training paradigms prioritize all data uniformly and train memory samples through multiple rounds, which would result in overfitting old tasks and pronounced bias towards new tasks because of the imbalances of the replay set. To handle the problem, we introduce the DecouPled CRE (DP-CRE) framework that decouples the process of prior information preservation and new knowledge acquisition. This framework examines alterations in the embedding space as new relation classes emerge, distinctly managing the preservation and acquisition of knowledge. Extensive experiments show that DP-CRE significantly outperforms other CRE baselines across two datasets.

Generative models are increasingly being used in various applications, such as text generation, commonsense reasoning, and question-answering. To be effective globally, these models must be aware of and account for local socio-cultural contexts, making it necessary to have benchmarks to evaluate the models for their cultural familiarity. Since the training data for LLMs is web-based and the Web is limited in its representation of information, it does not capture knowledge present within communities that are not on the Web. Thus, these models exacerbate the inequities, semantic misalignment, and stereotypes from the Web. There has been a growing call for community-centered participatory research methods in NLP. In this work, we respond to this call by using participatory research methods to introduce DOSA, the first community-generated Dataset of 615 Social Artifacts, by engaging with 260 participants from 19 different Indian geographic subcultures. We use a gamified framework that relies on collective sensemaking to collect the names and descriptions of these artifacts such that the descriptions semantically align with the shared sensibilities of the individuals from those cultures. Next, we benchmark four popular LLMs and find that they show significant variation across regional sub-cultures in their ability to infer the artifacts.

Definition modelling (DM) is the task of automatically generating a dictionary definition of a specific word. Computational systems that are capable of DM can have numerous applications benefiting a wide range of audiences. As DM is considered a supervised natural language generation problem, these systems require large annotated datasets to train the machine learning (ML) models. Several DM datasets have been released for English and other high-resource languages. While Portuguese is considered a mid/high-resource language in most natural language processing tasks and is spoken by more than 200 million native speakers, there is no DM dataset available for Portuguese. In this research, we fill this gap by introducing DORE; the first dataset for Definition MOdelling for PoRtuguEse containing more than 100,000 definitions. We also evaluate several deep learning based DM models on DORE and report the results. The dataset and the findings of this paper will facilitate research and study of Portuguese in wider contexts.

We propose DOC-RAG - Domain-distributed Co-occurrence Retrieval Augmentation for ASR language model personalization aiming to improve the automatic speech recognition of rare word patterns in unseen domains. Our approach involves contrastively training a document retrieval module to rank external knowledge domains based on their semantic similarity with respect to the input query. We further use n-gram co-occurrence distribution to recognize rare word patterns associated with specific domains. We aggregate the next word probability distribution based on the relative importance of different domains. Extensive experiments on three user-specific speech-to-text tasks for meetings, TED talks, and financial earnings calls show that DOC-RAG significantly outperforms strong baselines with an 8-15% improvement in terms of perplexity and a 4-7% reduction in terms of Word Error Rates in various settings.

Argument structure learning (ASL) entails predicting relations between arguments. Because it can structure a document to facilitate its understanding, it has been widely applied in many fields (medical, commercial, and scientific domains). Despite its broad utilization, ASL remains a challenging task because it involves examining the complex relationships between the sentences in a potentially unstructured discourse. To resolve this problem, we have developed a simple yet effective approach called Dual-tower Multi-scale cOnvolution neural Network (DMON) for the ASL task. Specifically, we organize arguments into a relationship matrix that together with the argument embeddings forms a relationship tensor and design a mechanism to capture relations with contextual arguments. Experimental results on three different-domain argument mining datasets demonstrate that our framework outperforms state-of-the-art models. We will release the code after paper acceptance.

This paper presents DISRPT, a multilingual, multi-domain, and cross-framework benchmark dataset for discourse processing, covering the tasks of discourse unit segmentation, connective identification, and relation classification. DISRPT includes 13 languages, with data from 24 corpora covering about 4 millions tokens and around 250,000 discourse relation instances from 4 discourse frameworks: RST, SDRT, PDTB, and Discourse Dependencies. We present an overview of the data, its development across three NLP shared tasks on discourse processing carried out in the past five years, and the latest modifications and added extensions. We also carry out an evaluation of state-of-the-art multilingual systems trained on the data for each task, showing plateau performance on segmentation, but important room for improvement for connective identification and relation classification. The DISRPT benchmark employs a unified format that we make available on GitHub and HuggingFace in order to encourage future work on discourse processing across languages, domains, and frameworks.

The method of training language models based on domain datasets has obtained significant achievements in the task of generating scientific paper abstracts. However, such models face problems of generalization and expensive training costs. The use of large language models (LLMs) to solve the task of generating paper abstracts saves the cost of model training. However, due to the hallucination problem of LLM, it is often necessary to improve the reliability of the results through multi-round query prompt approach such as Graph of Thoughts (GoT), which also brings additional reasoning costs. In this paper, we propose a Dynamic Graph of Thought (DGoT). It not only inherits the advantages of the existing GoT prompt approach, but also dynamically adjust the graph structure according to data characteristics while reducing model reasoning cost. Experimental results show that our method’s cost-effectiveness in abstract generation tasks is only 43.7% to 56.4% of other multi-round query prompt approaches. Our code is available at https://github.com/JayceNing/DGoT.

We present the acquisition process and the data of DGS-Fabeln-1, a parallel corpus of German text and videos containing German fairy tales interpreted into the German Sign Language (DGS) by a native DGS signer. The corpus contains 573 segments of videos with a total duration of 1 hour and 32 minutes, corresponding with 1428 written sentences. It is the first corpus of semi-naturally expressed DGS that has been filmed from 7 angles, and one of the few sign language (SL) corpora globally which have been filmed from more than 3 angles and where the listener has been simultaneously filmed. The corpus aims at aiding research at SL linguistics, SL machine translation and affective computing, and is freely available for research purposes at the following address: https://doi.org/10.5281/zenodo.10822097.