Sudipta Kar

Sudipta Kar, Giuseppe Castellucci, Simone Filice et al. · amazon-science

Multi-Task Learning (MTL) is widely-accepted in Natural Language Processing as a standard technique for learning multiple related tasks in one model. Training an MTL model requires having the training data for all tasks available at the same time. As systems usually evolve over time, (e.g., to support new functionalities), adding a new task to an existing MTL model usually requires retraining the model from scratch on all the tasks and this can be time-consuming and computationally expensive. Moreover, in some scenarios, the data used to train the original training may be no longer available, for example, due to storage or privacy concerns. In this paper, we approach the problem of incrementally expanding MTL models' capability to solve new tasks over time by distilling the knowledge of an already trained model on n tasks into a new one for solving n+1 tasks. To avoid catastrophic forgetting, we propose to exploit unlabeled data from the same distributions of the old tasks. Our experiments on publicly available benchmarks show that such a technique dramatically benefits the distillation by preserving the already acquired knowledge (i.e., preventing up to 20% performance drops on old tasks) while obtaining good performance on the incrementally added tasks. Further, we also show that our approach is beneficial in practical settings by using data from a leading voice assistant.

Shervin Malmasi, Anjie Fang, Besnik Fetahu et al.

We present MultiCoNER, a large multilingual dataset for Named Entity Recognition that covers 3 domains (Wiki sentences, questions, and search queries) across 11 languages, as well as multilingual and code-mixing subsets. This dataset is designed to represent contemporary challenges in NER, including low-context scenarios (short and uncased text), syntactically complex entities like movie titles, and long-tail entity distributions. The 26M token dataset is compiled from public resources using techniques such as heuristic-based sentence sampling, template extraction and slotting, and machine translation. We applied two NER models on our dataset: a baseline XLM-RoBERTa model, and a state-of-the-art GEMNET model that leverages gazetteers. The baseline achieves moderate performance (macro-F1=54%), highlighting the difficulty of our data. GEMNET, which uses gazetteers, improvement significantly (average improvement of macro-F1=+30%). MultiCoNER poses challenges even for large pre-trained language models, and we believe that it can help further research in building robust NER systems. MultiCoNER is publicly available at https://registry.opendata.aws/multiconer/ and we hope that this resource will help advance research in various aspects of NER.

Besnik Fetahu, Zhiyu Chen, Sudipta Kar et al.

We present MULTICONER V2, a dataset for fine-grained Named Entity Recognition covering 33 entity classes across 12 languages, in both monolingual and multilingual settings. This dataset aims to tackle the following practical challenges in NER: (i) effective handling of fine-grained classes that include complex entities like movie titles, and (ii) performance degradation due to noise generated from typing mistakes or OCR errors. The dataset is compiled from open resources like Wikipedia and Wikidata, and is publicly available. Evaluation based on the XLM-RoBERTa baseline highlights the unique challenges posed by MULTICONER V2: (i) the fine-grained taxonomy is challenging, where the scores are low with macro-F1=0.63 (across all languages), and (ii) the corruption strategy significantly impairs performance, with entity corruption resulting in 9% lower performance relative to non-entity corruptions across all languages. This highlights the greater impact of entity noise in contrast to context noise.

Chris Richardson, Yao Zhang, Kellen Gillespie et al.

Personalization, the ability to tailor a system to individual users, is an essential factor in user experience with natural language processing (NLP) systems. With the emergence of Large Language Models (LLMs), a key question is how to leverage these models to better personalize user experiences. To personalize a language model's output, a straightforward approach is to incorporate past user data into the language model prompt, but this approach can result in lengthy inputs exceeding limitations on input length and incurring latency and cost issues. Existing approaches tackle such challenges by selectively extracting relevant user data (i.e. selective retrieval) to construct a prompt for downstream tasks. However, retrieval-based methods are limited by potential information loss, lack of more profound user understanding, and cold-start challenges. To overcome these limitations, we propose a novel summary-augmented approach by extending retrieval-augmented personalization with task-aware user summaries generated by LLMs. The summaries can be generated and stored offline, enabling real-world systems with runtime constraints like voice assistants to leverage the power of LLMs. Experiments show our method with 75% less of retrieved user data is on-par or outperforms retrieval augmentation on most tasks in the LaMP personalization benchmark. We demonstrate that offline summarization via LLMs and runtime retrieval enables better performance for personalization on a range of tasks under practical constraints.

Christopher Richardson, Sudipta Kar, Anjishnu Kumar et al.

Open domain conversational agents can answer a broad range of targeted queries. However, the sequential nature of interaction with these systems makes knowledge exploration a lengthy task which burdens the user with asking a chain of well phrased questions. In this paper, we present a retrieval based system and associated dataset for predicting the next questions that the user might have. Such a system can proactively assist users in knowledge exploration leading to a more engaging dialog. The retrieval system is trained on a dataset which contains ~14K multi-turn information-seeking conversations with a valid follow-up question and a set of invalid candidates. The invalid candidates are generated to simulate various syntactic and semantic confounders such as paraphrases, partial entity match, irrelevant entity, and ASR errors. We use confounder specific techniques to simulate these negative examples on the OR-QuAC dataset and develop a dataset called the Follow-up Query Bank (FQ-Bank). Then, we train ranking models on FQ-Bank and present results comparing supervised and unsupervised approaches. The results suggest that we can retrieve the valid follow-ups by ranking them in higher positions compared to confounders, but further knowledge grounding can improve ranking performance.

Zabir Al Nazi, Shubhashis Roy Dipta, Sudipta Kar

Chain-of-Thought (CoT) prompting is widely adopted for mathematical problem solving, including in low-resource languages, yet its behavior under irrelevant context remains underexplored. To systematically study this challenge, we introduce DISTRACTMATH-BN, a Bangla benchmark that augments MGSM and MSVAMP with semantically coherent but computationally irrelevant information. Evaluating seven models ranging from 3B to 12B parameters, we observe substantial performance degradation under distractors: standard models drop by up to 41 points, while reasoning-specialized models decline by 14 to 20 points despite consuming five times more tokens. We propose †DAGGER, which reformulates mathematical problem solving as executable computational graph generation with explicit modeling of distractor nodes. Fine-tuning Gemma-3 models using supervised fine-tuning followed by Group Relative Policy Optimization achieves comparable weighted accuracy on augmented benchmarks while using 89 percent fewer tokens than reasoning models. Importantly, this robustness emerges without explicit training on distractor-augmented examples. Our results suggest that enforcing structured intermediate representations improves robustness and inference efficiency in mathematical reasoning compared to free-form approaches, particularly in noisy, low-resource settings.

Genta Indra Winata, David Anugraha, Emmy Liu et al. · amazon-science

High-quality datasets are fundamental to training and evaluating machine learning models, yet their creation-especially with accurate human annotations-remains a significant challenge. Many dataset paper submissions lack originality, diversity, or rigorous quality control, and these shortcomings are often overlooked during peer review. Submissions also frequently omit essential details about dataset construction and properties. While existing tools such as datasheets aim to promote transparency, they are largely descriptive and do not provide standardized, measurable methods for evaluating data quality. Similarly, metadata requirements at conferences promote accountability but are inconsistently enforced. To address these limitations, this position paper advocates for the integration of systematic, rubric-based evaluation metrics into the dataset review process-particularly as submission volumes continue to grow. We also explore scalable, cost-effective methods for synthetic data generation, including dedicated tools and LLM-as-a-judge approaches, to support more efficient evaluation. As a call to action, we introduce DataRubrics, a structured framework for assessing the quality of both human- and model-generated datasets. Leveraging recent advances in LLM-based evaluation, DataRubrics offers a reproducible, scalable, and actionable solution for dataset quality assessment, enabling both authors and reviewers to uphold higher standards in data-centric research. We also release code to support reproducibility of LLM-based evaluations at https://github.com/datarubrics/datarubrics.

Shirley Anugrah Hayati, Taehee Jung, Tristan Bodding-Long et al.

Fine-tuning large language models (LLMs) with a collection of large and diverse instructions has improved the model's generalization to different tasks, even for unseen tasks. However, most existing instruction datasets include only single instructions, and they struggle to follow complex instructions composed of multiple subtasks. In this work, we propose a novel concept of compositional instructions called chain-of-instructions (CoI), where the output of one instruction becomes an input for the next like a chain. Unlike the conventional practice of solving single instruction tasks, our proposed method encourages a model to solve each subtask step by step until the final answer is reached. CoI-tuning (i.e., fine-tuning with CoI instructions) improves the model's ability to handle instructions composed of multiple subtasks as well as unseen composite tasks such as multilingual summarization. Overall, our study find that simple CoI tuning of existing instruction data can provide consistent generalization to solve more complex, unseen, and longer chains of instructions.

Besnik Fetahu, Sudipta Kar, Zhiyu Chen et al.

We present the findings of SemEval-2023 Task 2 on Fine-grained Multilingual Named Entity Recognition (MultiCoNER 2). Divided into 13 tracks, the task focused on methods to identify complex fine-grained named entities (like WRITTENWORK, VEHICLE, MUSICALGRP) across 12 languages, in both monolingual and multilingual scenarios, as well as noisy settings. The task used the MultiCoNER V2 dataset, composed of 2.2 million instances in Bangla, Chinese, English, Farsi, French, German, Hindi, Italian., Portuguese, Spanish, Swedish, and Ukrainian. MultiCoNER 2 was one of the most popular tasks of SemEval-2023. It attracted 842 submissions from 47 teams, and 34 teams submitted system papers. Results showed that complex entity types such as media titles and product names were the most challenging. Methods fusing external knowledge into transformer models achieved the best performance, and the largest gains were on the Creative Work and Group classes, which are still challenging even with external knowledge. Some fine-grained classes proved to be more challenging than others, such as SCIENTIST, ARTWORK, and PRIVATECORP. We also observed that noisy data has a significant impact on model performance, with an average drop of 10% on the noisy subset. The task highlights the need for future research on improving NER robustness on noisy data containing complex entities.

Parth Patwa, Gustavo Aguilar, Sudipta Kar et al.

In this paper, we present the results of the SemEval-2020 Task 9 on Sentiment Analysis of Code-Mixed Tweets (SentiMix 2020). We also release and describe our Hinglish (Hindi-English) and Spanglish (Spanish-English) corpora annotated with word-level language identification and sentence-level sentiment labels. These corpora are comprised of 20K and 19K examples, respectively. The sentiment labels are - Positive, Negative, and Neutral. SentiMix attracted 89 submissions in total including 61 teams that participated in the Hinglish contest and 28 submitted systems to the Spanglish competition. The best performance achieved was 75.0% F1 score for Hinglish and 80.6% F1 for Spanglish. We observe that BERT-like models and ensemble methods are the most common and successful approaches among the participants.

Gustavo Aguilar, Sudipta Kar, Thamar Solorio

Recent trends in NLP research have raised an interest in linguistic code-switching (CS); modern approaches have been proposed to solve a wide range of NLP tasks on multiple language pairs. Unfortunately, these proposed methods are hardly generalizable to different code-switched languages. In addition, it is unclear whether a model architecture is applicable for a different task while still being compatible with the code-switching setting. This is mainly because of the lack of a centralized benchmark and the sparse corpora that researchers employ based on their specific needs and interests. To facilitate research in this direction, we propose a centralized benchmark for Linguistic Code-switching Evaluation (LinCE) that combines ten corpora covering four different code-switched language pairs (i.e., Spanish-English, Nepali-English, Hindi-English, and Modern Standard Arabic-Egyptian Arabic) and four tasks (i.e., language identification, named entity recognition, part-of-speech tagging, and sentiment analysis). As part of the benchmark centralization effort, we provide an online platform at ritual.uh.edu/lince, where researchers can submit their results while comparing with others in real-time. In addition, we provide the scores of different popular models, including LSTM, ELMo, and multilingual BERT so that the NLP community can compare against state-of-the-art systems. LinCE is a continuous effort, and we will expand it with more low-resource languages and tasks.

Md Zobaer Hossain, Md Ashraful Rahman, Md Saiful Islam et al.

Observing the damages that can be done by the rapid propagation of fake news in various sectors like politics and finance, automatic identification of fake news using linguistic analysis has drawn the attention of the research community. However, such methods are largely being developed for English where low resource languages remain out of the focus. But the risks spawned by fake and manipulative news are not confined by languages. In this work, we propose an annotated dataset of ~50K news that can be used for building automated fake news detection systems for a low resource language like Bangla. Additionally, we provide an analysis of the dataset and develop a benchmark system with state of the art NLP techniques to identify Bangla fake news. To create this system, we explore traditional linguistic features and neural network based methods. We expect this dataset will be a valuable resource for building technologies to prevent the spreading of fake news and contribute in research with low resource languages.

Niloofar Safi Samghabadi, Afsheen Hatami, Mahsa Shafaei et al.

In recent years, abusive behavior has become a serious issue in online social networks. In this paper, we present a new corpus from a semi-anonymous social media platform, which contains the instances of offensive and neutral classes. We introduce a single deep neural architecture that considers both local and sequential information from the text in order to detect abusive language. Along with this model, we introduce a new attention mechanism called emotion-aware attention. This mechanism utilizes the emotions behind the text to find the most important words within that text. We experiment with this model on our dataset and later present the analysis. Additionally, we evaluate our proposed method on different corpora and show new state-of-the-art results with respect to offensive language detection.

Sudipta Kar, Gustavo Aguilar, Mirella Lapata et al.

This paper considers the problem of characterizing stories by inferring properties such as theme and style using written synopses and reviews of movies. We experiment with a multi-label dataset of movie synopses and a tagset representing various attributes of stories (e.g., genre, type of events). Our proposed multi-view model encodes the synopses and reviews using hierarchical attention and shows improvement over methods that only use synopses. Finally, we demonstrate how can we take advantage of such a model to extract a complementary set of story-attributes from reviews without direct supervision. We have made our dataset and source code publicly available at https://ritual.uh.edu/ multiview-tag-2020.

Mahsa Shafaei, Niloofar Safi Samghabadi, Sudipta Kar et al.

The film culture has grown tremendously in recent years. The large number of streaming services put films as one of the most convenient forms of entertainment in today's world. Films can help us learn and inspire societal change. But they can also negatively affect viewers. In this paper, our goal is to predict the suitability of the movie content for children and young adults based on scripts. The criterion that we use to measure suitability is the MPAA rating that is specifically designed for this purpose. We propose an RNN based architecture with attention that jointly models the genre and the emotions in the script to predict the MPAA rating. We achieve 78% weighted F1-score for the classification model that outperforms the traditional machine learning method by 6%.

Sudipta Kar, Suraj Maharjan, Thamar Solorio

Folksonomy of movies covers a wide range of heterogeneous information about movies, like the genre, plot structure, visual experiences, soundtracks, metadata, and emotional experiences from watching a movie. Being able to automatically generate or predict tags for movies can help recommendation engines improve retrieval of similar movies, and help viewers know what to expect from a movie in advance. In this work, we explore the problem of creating tags for movies from plot synopses. We propose a novel neural network model that merges information from synopses and emotion flows throughout the plots to predict a set of tags for movies. We compare our system with multiple baselines and found that the addition of emotion flows boosts the performance of the network by learning ~18\% more tags than a traditional machine learning system.

Niloofar Safi Samghabadi, Deepthi Mave, Sudipta Kar et al.

This paper presents our system for "TRAC 2018 Shared Task on Aggression Identification". Our best systems for the English dataset use a combination of lexical and semantic features. However, for Hindi data using only lexical features gave us the best results. We obtained weighted F1- measures of 0.5921 for the English Facebook task (ranked 12th), 0.5663 for the English Social Media task (ranked 6th), 0.6292 for the Hindi Facebook task (ranked 1st), and 0.4853 for the Hindi Social Media task (ranked 2nd).

Marc Franco-Salvador, Sudipta Kar, Thamar Solorio et al.

In this work we describe the system built for the three English subtasks of the SemEval 2016 Task 3 by the Department of Computer Science of the University of Houston (UH) and the Pattern Recognition and Human Language Technology (PRHLT) research center - Universitat Polit`ecnica de Val`encia: UH-PRHLT. Our system represents instances by using both lexical and semantic-based similarity measures between text pairs. Our semantic features include the use of distributed representations of words, knowledge graphs generated with the BabelNet multilingual semantic network, and the FrameNet lexical database. Experimental results outperform the random and Google search engine baselines in the three English subtasks. Our approach obtained the highest results of subtask B compared to the other task participants.

Suraj Maharjan, Sudipta Kar, Manuel Montes-y-Gomez et al.

Books have the power to make us feel happiness, sadness, pain, surprise, or sorrow. An author's dexterity in the use of these emotions captivates readers and makes it difficult for them to put the book down. In this paper, we model the flow of emotions over a book using recurrent neural networks and quantify its usefulness in predicting success in books. We obtained the best weighted F1-score of 69% for predicting books' success in a multitask setting (simultaneously predicting success and genre of books).

Sudipta Kar, Suraj Maharjan, A. Pastor López-Monroy et al.

Social tagging of movies reveals a wide range of heterogeneous information about movies, like the genre, plot structure, soundtracks, metadata, visual and emotional experiences. Such information can be valuable in building automatic systems to create tags for movies. Automatic tagging systems can help recommendation engines to improve the retrieval of similar movies as well as help viewers to know what to expect from a movie in advance. In this paper, we set out to the task of collecting a corpus of movie plot synopses and tags. We describe a methodology that enabled us to build a fine-grained set of around 70 tags exposing heterogeneous characteristics of movie plots and the multi-label associations of these tags with some 14K movie plot synopses. We investigate how these tags correlate with movies and the flow of emotions throughout different types of movies. Finally, we use this corpus to explore the feasibility of inferring tags from plot synopses. We expect the corpus will be useful in other tasks where analysis of narratives is relevant.