BENGALI

Table Question Answering for Low-resourced Indic Languages (2024.emnlp-main)

Challenge:	Existing studies on the effectiveness of different architectures for modeling low resource languages are limited.
Approach:	They propose a trainable memory efficient CNN architecture for Bengali and Hindi . they propose two learnable convolutional sub-models that are end to end trainable .
Outcome:	The proposed model outperforms pretrained BERT models on Bengali and Hindi with 16X less parameters and achieves much better performance than SOTA LSTMs on multiple real-world datasets.

Copied to clipboard

Vaishali Pal, Evangelos Kanoulas, Andrew Yates, Maarten Rijke

Structured Data in NLP

Challenge:	TableQA is the task of answering questions over tables of structured information, returning individual cells or tables as output.
Approach:	They propose a fully automatic large-scale tableQA data generation process for low-resource languages with limited budget.
Outcome:	The proposed method outperforms state-of-the-art LLMs on two Indic languages with no tableQA datasets and models on different aspects including mathematical reasoning capabilities and zero-shot cross-lingual transfer.

Emotion Classification in a Resource Constrained Language Using Transformer-based Approach (2021.naacl-srw)

Copied to clipboard

Avishek Das, Omar Sharif, Mohammed Moshiul Hoque, Iqbal H. Sarker

How Transferable are Attribute Controllers on Pretrained Multilingual Translation Models? (2024.eacl-long)

Challenge:	Existing methods to classify Bengali text into six basic emotions are infancy for resource-constrained languages like English, Arabic, Chinese and French.
Approach:	They propose a transformer-based technique to classify Bengali text into one of the six basic emotions: anger, fear, disgust, sadness, joy, and surprise.
Outcome:	The proposed technique outperforms all other techniques by achieving highest weighted f_1-score on the test data.

Copied to clipboard

Danni Liu, Jan Niehues

Hate Speech and Offensive Language Detection in Bengali (2022.aacl-main)

Challenge:	Pretrained multilingual translation models with massive coverage are becoming of the backbone of many translation systems.
Approach:	They propose to use a gradient-based inference-time controller to control a pretrained multilingual model by using a model with attribute annotations.
Outcome:	The proposed model performs well on pretrained multilingual models and is attribute- rather than language-specific.

Copied to clipboard

Mithun Das, Somnath Banerjee, Punyajoy Saha, Animesh Mukherjee

Information Retrieval

Challenge:	Existing research on hate speech detection in English does not cover low-resource languages like Bengali.
Approach:	They develop an annotated dataset of 10K Bengali posts consisting of 5K actual and 5K Romanized Bengali tweets.
Outcome:	The proposed model outperforms other models on training actual and romanized datasets by interpreting the semantic expressions better.

Cross-Lingual Open-Domain Question Answering with Answer Sentence Generation (2022.aacl-main)

Copied to clipboard

Benjamin Muller, Luca Soldaini, Rik Koncel-Kedziorski, Eric Lind, Alessandro Moschitti

Natural Language Interfaces

Challenge:	Open-Domain Generative Question Answering has achieved impressive performance in English . combining document-level retrieval with answer generation can generate complete sentences .
Approach:	They propose an open-domain approach that combines document retrieval with answer generation to generate complete sentences in English . they propose a cross-lingual generative model that exploits passages written in multiple languages .
Outcome:	The proposed model outperforms answer sentence selection baselines for all 5 languages and monolingual pipelines for three out of five languages.

Robust Neural Machine Translation for Abugidas by Glyph Perturbation (2024.eacl-short)

Copied to clipboard

Hour Kaing, Chenchen Ding, Hideki Tanaka, Masao Utiyama

Confidence as a Tie-Breaker: Reassessing Multilingual Hedging Bias in LLM-as-a-Judge Evaluation (2026.acl-srw)

Challenge:	Neural machine translation systems are vulnerable when trained on limited data.
Approach:	They propose to add noise to the training phase to increase robustness of NMT systems trained on limited data.
Outcome:	The proposed training strategy overcomes noise and improves robustness for low-resource tasks for abugida glyphs.

Copied to clipboard

Rajashik Datta, Sanjan Baitalik

Evaluating Text Style Transfer Evaluation: Are There Any Reliable Metrics? (2025.naacl-srw)

Challenge:	LLM judges are often used to score generated answers, but their decisions may be affected by surface style rather than semantic correctness.
Approach:	They propose a benchmark to examine multilingual hedging effects in LLM evaluation . they find that when two answers are equally correct, the judge prefers the assertive answer .
Outcome:	The proposed benchmark shows that judge preferences are based on style and style . the results highlight that benchmark auditing is a key requirement for judge-bias research .

Copied to clipboard

Sourabrata Mukherjee, Atul Kr. Ojha, John Philip McCrae, Ondrej Dusek

Text Generation

Challenge:	Text style transfer (TST) is a multidimensional task requiring the assessment of style transfer accuracy, content preservation, and naturalness.
Approach:	They propose to use text style transfer metrics to evaluate outputs of text editors . they also investigate the potential of large language models as tools for TST evaluation .
Outcome:	The proposed methods provide better insights than existing metrics, the authors show . their meta-evaluation through correlation with hu-man judgments shows they are effective .

IndicIRSuite: Multilingual Dataset and Neural Information Models for Indian Languages (2024.acl-short)

Copied to clipboard

Saiful Haq, Ashutosh Sharma, Omar Khattab, Niyati Chhaya, Pushpak Bhattacharyya

GraDeT-HTR: A Resource-Efficient Bengali Handwritten Text Recognition System utilizing Grapheme-based Tokenizer and Decoder-only Transformer (2025.emnlp-demos)

Challenge:	IndicIRSuite is the first attempt at building large-scale Neural Information Retrieval resources for a large number of Indian languages.
Approach:	They introduce Neural Information Retrieval resources for 11 widely spoken Indian Languages from two major Indian language families.
Outcome:	Experiments show that Indic-ColBERT improves on INDIC-MARCO datasets for 11 languages, and that it can be used to improve IR for Indian languages.

Copied to clipboard

Md. Mahmudul Hasan, Ahmed Nesar Tahsin Choudhury, Mahmudul Hasan, Md Mosaddek Khan

BEnQA: A Question Answering Benchmark for Bengali and English (2024.findings-acl)

Challenge:	Bengali is the sixth most spoken language in the world, but handwritten text recognition systems for the language are underdeveloped.
Approach:	They propose a Bengali handwritten text recognition system that uses a decoder-only transformer to address the unique challenges of Bengali script.
Outcome:	The proposed system significantly improves on existing tokenizers on Bengali script.

Copied to clipboard

Sheikh Shafayat, H Hasan, Minhajur Mahim, Rifki Putri, James Thorne, Alice Oh

Low-Resource NLP

Challenge:	a dataset of parallel Bengali and English exam questions is used to compare LLMs in low-resource languages.
Approach:	They introduce BEnQA, a dataset comprising parallel Bengali and English exam questions . they benchmark several Large Language Models with their parallel dataset and observe performance disparity .
Outcome:	The proposed dataset consists of 5K questions covering several subjects in science . the authors find that the models perform poorly in Bengali and English .

Too Late to Train, Too Early To Use? A Study on Necessity and Viability of Low-Resource Bengali LLMs (2025.coling-main)

Copied to clipboard

Tamzeed Mahfuz, Satak Kumar Dey, Ruwad Naswan, Hasnaen Adil, Khondker Salman Sayeed, Haz Sameen Shahgir

Joint Transformer/RNN Architecture for Gesture Typing in Indic Languages (2020.coling-main)

Challenge:	a new generation of English-oriented Large Language Models significantly outperforms older LLMs on low-resource languages.
Approach:	They compare Bengali-oriented LLMs with open-weight and closed-source LLM models . they conclude that there is a need for a Bengali model, but lacks high-quality pretraining data .
Outcome:	The proposed model outperforms existing models on Bengali on low-resource languages . the results highlight biases in machine-translated datasets used for Bengali NLP tasks .

Copied to clipboard

Emil Biju, Anirudh Sriram, Mitesh M. Khapra, Pratyush Kumar

Bidirectional Reasoning Supervision for Multilingual Financial Decision Making (2025.emnlp-industry)

Challenge:	Gesture typing is a method of typing words on a touch-based keyboard by drawing a continuous trace passing through the relevant keys.
Approach:	They propose a keyboard that supports gesture typing in Indic languages by drawing a continuous trace over the keyboard and the finger needs to be lifted only once a word is completed.
Outcome:	The proposed model performs path decoding, transliteration and transliterations correction.

Copied to clipboard

Muhammad Rafsan Kabir, Jawad Ibn Ahad, Robin Krambroeckers, Silvia Ahmed, M M Lutfe Elahi, Nabeel Mohammed, Shafin Rahman

Information Retrieval

Challenge:	Large Language Models have been used for sentiment analysis, machine translation, and question answering, but their effectiveness in the multilingual financial domain remains unknown.
Approach:	They propose a fine-tuning approach that integrates positive and negative rationales alongside classification labels.
Outcome:	The proposed approach outperforms existing methods across English, Hindi, Bengali, and Telugu, and is suitable for industry applications.

Low-Resource Counterspeech Generation for Indic Languages: The Case of Bengali and Hindi (2024.findings-eacl)

Copied to clipboard

Mithun Das, Saurabh Pandey, Shivansh Sethi, Punyajoy Saha, Animesh Mukherjee

Harnessing Cross-lingual Features to Improve Cognate Detection for Low-resource Languages (2020.coling-main)

Challenge:	NLP is a technique that generates counterspeech that “counters” the vicious tone of online abuse and dilutes/ameliorates their rippling effect over the social network.
Approach:	They propose to use neural architectures to generate counterspeech that can "counter" the vicious tone of online abuse and dilute/ameliorate their rippling effect over the social network.
Outcome:	The proposed model can generate counterspeech in monolingual setups and is more transferable when languages belong to the same language family.

Copied to clipboard

Diptesh Kanojia, Raj Dabre, Shubham Dewangan, Pushpak Bhattacharyya, Gholamreza Haffari, Malhar Kulkarni

Twitter corpus of Resource-Scarce Languages for Sentiment Analysis and Multilingual Emoji Prediction (C18-1)

Challenge:	a study of 14 Indian languages shows that cognates can be detected by word embeddings . cognates are variants of the same lexical form across languages .
Approach:	They propose to use cross-lingual word embeddings to detect cognates among 14 Indian languages . they then evaluate the impact of their method on neural machine translation .
Outcome:	The proposed method improves on a dataset of 12 Indian languages . it also improves quality of the extracted cognates by up to 2.76 BLEU .

Copied to clipboard

Nurendra Choudhary, Rajat Singh, Vijjini Anvesh Rao, Manish Shrivastava

Do LLM hallucination detectors suffer from low-resource effect? (2026.eacl-long)

Challenge:	a majority of research studies on twitter focus on English tweets, despite the fact that English dominates the mix of languages.
Approach:	They leverage social media platforms such as twitter for developing corpus across multiple languages . they use tweets to collect data for sentiment analysis and emoji prediction .
Outcome:	The proposed method is applicable for resource-scarce languages provided speakers of that particular language are active users on social media platforms.

Copied to clipboard

Debtanu Datta, Mohan Kishore Chilukuri, Yash Kumar, Saptarshi Ghosh, Muhammad Bilal Zafar

Intent Identification and Entity Extraction for Healthcare Queries in Indic Languages (2023.findings-eacl)

Challenge:	a long line of work suggests that LLMs face issues along both dimensions .
Approach:	They investigate hallucination detectors' failure modes and their effects on the task accuracy of four LLMs and three halluciner detectors.
Outcome:	The models show impressive performance in high-resource languages like English but the performance degrades significantly in low-resourced languages like Bengali.

Copied to clipboard

Ankan Mullick, Ishani Mondal, Sourjyadip Ray, Raghav R, G Chaitanya, Pawan Goyal

A Multimodal Framework to Detect Target Aware Aggression in Memes (2024.eacl-long)

Challenge:	Currently, there is a lack of data and technology for resource-poor languages in developing countries like India.
Approach:	They propose to use two different datasets to analyze query intents and entities in healthcare.
Outcome:	The proposed model is useful to identify query intents and entities in real-world scenarios.

Copied to clipboard

Shawly Ahsan, Eftekhar Hossain, Omar Sharif, Avishek Das, Mohammed Moshiul Hoque, M. Dewan

MemoSen: A Multimodal Dataset for Sentiment Analysis of Memes (2022.lrec-1)

Challenge:	Recent research on memes’ detrimental facets is skewed towards high-resource languages, such as Bengali.
Approach:	They propose a dataset MIMOSA that annotates annotated memes across five aggression target categories in Bengali and propose 'Multimodal Attentive Fusion' to detect aggression targets.
Outcome:	The proposed method outperforms state-of-the-art methods in Bengali and in low-resource languages.

Copied to clipboard

Eftekhar Hossain, Omar Sharif, Mohammed Moshiul Hoque

BanglaAutoKG: Automatic Bangla Knowledge Graph Construction with Semantic Neural Graph Filtering (2024.lrec-main)

Challenge:	Recent studies on sentiment analysis of memes have focused on English, but there is a significant barrier to performing multimodal sentiment analysis research in resource-constrained languages like Bengali.
Approach:	They propose to use a Bengali dataset to perform multimodal sentiment analysis in low resource languages.
Outcome:	The proposed dataset for Bengali contains 4417 memes with three annotated labels positive, negative, and neutral.

Copied to clipboard

Azmine Toushik Wasi, Taki Hasan Rafi, Raima Islam, Dong-Kyu Chae

Structured Data in NLP

Challenge:	Bangla is underrepresented in KGs due to lack of comprehensive datasets, encoders, NER models, part-of-speech taggers, and lemmatizers.
Approach:	Bangla is underrepresented in KGs due to lack of comprehensive datasets, encoders, NER models, part-of-speech taggers, and lemmatizers. authors propose a framework that can automatically construct Bengali KG from any Bangla text.
Outcome:	The proposed framework can automatically construct Bengali KGs from any Bangla text.

Not Low-Resource Anymore: Aligner Ensembling, Batch Filtering, and New Datasets for Bengali-English Machine Translation (2020.emnlp-main)

Copied to clipboard

Tahmid Hasan, Abhik Bhattacharjee, Kazi Samin, Masum Hasan, Madhusudan Basak, M. Sohel Rahman, Rifat Shahriyar

Machine Translation

Challenge:	despite being the seventh most widely spoken language, Bengali has received little attention in machine translation due to being low in resources.
Approach:	They propose a customized sentence segmenter for Bengali and two new methods for parallel corpus creation on low-resource setups.
Outcome:	The proposed method improves Bengali-English parallel corpus by 9 BLEU over previous approaches . the results will pave the way for future research on Bengali and other low-resource languages .

Towards Fair and Efficient De-identification: Quantifying the Efficiency and Generalizability of De-identification Approaches (2026.findings-eacl)

Copied to clipboard

Noopur Zambare, Kiana Aghakasiri, Carissa Lin, Carrie Ye, J Ross Mitchell, Mohamed Abdalla

Unsupervised Abstractive Summarization of Bengali Text Documents (2021.eacl-main)

Challenge:	a recent study has not examined their generalizability between formats, cultures, and genders.
Approach:	They evaluate large language models (LLMs) and small LLMs at clinical de-identification . they show that smaller models achieve comparable performance while substantially reducing inference cost .
Outcome:	The proposed models outperform larger models in de-identification tasks with limited data . the models can be fine-tuned with limited datasets to outperformed larger models .

Copied to clipboard

Radia Rayan Chowdhury, Mir Tafseer Nayeem, Tahsin Tasnim Mim, Md. Saifur Rahman Chowdhury, Taufiqul Jannat

A Study on Scaling Up Multilingual News Framing Analysis (2024.findings-naacl)

Challenge:	Abstractive summarization systems are difficult to perform due to the unavailability of the parallel data for low-resource languages like Bengali.
Approach:	They propose a graph-based unsupervised abstractive summarization system in Bengali text documents that requires only a Part-Of-Speech (POS) tagger and a pre-trained language model trained on Bengali texts.
Outcome:	The proposed system outperforms baselines without human-annotated reference summaries on a human-random dataset with Bengali text.

Copied to clipboard

Syeda Sabrina Akter, Antonios Anastasopoulos

SD-QA: Spoken Dialectal Question Answering for the Real World (2021.findings-emnlp)

Challenge:	Existing studies on media framing have focused on English only data, leaving a gap in research concerning multilingual contexts.
Approach:	They propose to use crowd-sourced datasets to automate framing analysis by automating translation and annotation.
Outcome:	The proposed system improves on existing models in Bengali and Portuguese . the proposed system can train on a crowd-sourced dataset in 12 languages .

Copied to clipboard

Fahim Faisal, Sharlina Keshava, Md Mahfuz Ibn Alam, Antonios Anastasopoulos

Claim Matching Beyond English to Scale Global Fact-Checking (2021.acl-long)

Challenge:	Existing QA benchmarks do not account for errors that speech recognition models might introduce . evaluating production-ready QA systems on data that is not representative of real-world inputs is problematic .
Approach:	They construct a multi-dialect, spoken QA benchmark on five languages with 68k audio prompts in 24 dialects from 255 speakers.
Outcome:	The proposed model is based on 68k audio prompts in 24 dialects from 255 speakers.

Copied to clipboard

Ashkan Kazemi, Kiran Garimella, Devin Gaffney, Scott Hale

Abstractive Summarization of Bengali Academic Videos Based on Audio Subtitles (2026.findings-eacl)

Challenge:	Existing methods to fact-check content are not scaled well in non-English contexts.
Approach:	They propose to use a WhatsApp tipline and public group message dataset to find pairs of textual messages containing claims that can be served with one fact-check.
Outcome:	The proposed model outperforms existing models in English, Hindi, and Tamil in all settings.

Copied to clipboard

Lamisa Bintee Mizan Deya, Farhatun Shama, Abdul Aziz, Md Kaykobad Reza, Md Shahidul Salim

Adapting Word Embeddings to New Languages with Morphological and Phonological Subword Representations (D18-1)

Challenge:	Existing methods for summarizing educational videos in Bengali are limited due to the rapid growth of educational video content.
Approach:	They propose an end-to-end pipeline for the abstractive summarization of Bengali videos . they fine-tuned the BanglaT5 model on a new benchmark dataset .
Outcome:	The proposed system preprocesses audio and converts speech to text using Google's Speech Recognition API.

Copied to clipboard

Aditi Chaudhary, Chunting Zhou, Lori Levin, Graham Neubig, David R. Mortensen, Jaime Carbonell

Syntactic Text Processing

Challenge:	Existing approaches to generalization to resource-rich languages are difficult . a recent study shows that word representations can be useful in low resource languages .
Approach:	They propose two approaches for improving generalization to low-resource languages by adapting continuous word representations using linguistically motivated subword units.
Outcome:	The proposed method improves generalization to low resource languages . it requires neither parallel corpora nor bilingual dictionaries and requires no parallel training .

Does Meta-learning Help mBERT for Few-shot Question Generation in a Cross-lingual Transfer Setting for Indic Languages? (2022.coling-1)

Copied to clipboard

Aniruddha Roy, Rupak Kumar Thakur, Isha Sharma, Ashim Gupta, Amrith Krishna, Sudeshna Sarkar, Pawan Goyal

Low-Resource NLP

Challenge:	Existing approaches to few-shot Question Generation (QG) are limited and require manual annotation.
Approach:	They propose to use multilingual BERT to perform few-shot question generation with cross-lingual transfer.
Outcome:	The proposed model improves in few-shot QG and human evaluation confirms it.

Challenge Dataset of Cognates and False Friend Pairs from Indian Languages (2020.lrec-1)

Copied to clipboard

Diptesh Kanojia, Malhar Kulkarni, Pushpak Bhattacharyya, Gholamreza Haffari

NERvous About My Health: Constructing a Bengali Medical Named Entity Recognition Dataset (2023.findings-emnlp)

Challenge:	Cognates are words that have a common etymological origin and can facilitate the Second Language Acquisition (SLA) however, they also pose a challenge to various NLP applications such as Machine Translation and Cross-lingual Sense Disambiguation.
Approach:	They create two cognate datasets for twelve Indian languages and use them to generate cognate sets.
Outcome:	The proposed datasets are curated using previously available baseline cognate detection approaches and evaluated with the help of lexicographers.

Copied to clipboard

Alvi Khan, Fida Kamal, Nuzhat Nower, Tasnim Ahmed, Sabbir Ahmed, Tareque Chowdhury

Zero-shot Disfluency Detection for Indian Languages (2022.coling-1)

Challenge:	Named Entity Recognition (NER) is used in a variety of downstream tasks in the biomedical domain, but is difficult when working with consumer health questions (CHQs).
Approach:	They propose to use a dataset to identify named entities in health-related texts in Bengali to address the scarcity of available data.
Outcome:	The proposed dataset captures the diverse range of linguistic styles and dialects used by native speakers from various regions in their day-to-day lives.

Copied to clipboard

Rohit Kundu, Preethi Jyothi, Pushpak Bhattacharyya

HIT - A Hierarchically Fused Deep Attention Network for Robust Code-mixed Language Representation (2021.findings-acl)

Challenge:	Disfluency correction models can help alleviate this problem, but the unavailability of labeled data in low-resource languages impairs progress.
Approach:	They propose to use a pretrained multilingual model to detect zero-shot disfluency in Indian languages.
Outcome:	The proposed model achieves F1 scores of 75 and higher on five disfluency types across four languages.

Copied to clipboard

Ayan Sengupta, Sourabh Kumar Bhattacharjee, Tanmoy Chakraborty, Md. Shad Akhtar

Semi-Automated Construction of Sense-Annotated Datasets for Practically Any Language (2025.coling-main)

Challenge:	linguistics and morphology of resource-short code-mixed texts remain a key challenge in text processing.
Approach:	They propose a hierarchical transformer-based framework that captures the semantic relationship among words and hierarchically learns sentencelevel semantics using a fused attention mechanism.
Outcome:	The proposed framework improves on one European and five Indic languages on four NLP tasks on eleven datasets.

Copied to clipboard

Jai Riley, Bradley M. Hauer, Nafisa Sadaf Hriti, Guoqing Luo, Amir Reza Mirzaei, Ali Rafiei, Hadi Sheikhi, Mahvash Siavashpour, Mohammad Tavakoli, Ning Shi, Grzegorz Kondrak

PARIKSHA: A Large-Scale Investigation of Human-LLM Evaluator Agreement on Multilingual and Multi-Cultural Data (2024.emnlp-main)

Challenge:	Word sense disambiguation is a widely studied NLP task of identifying the meaning of a word in context.
Approach:	They propose a method to create parallel sense-annotated datasets in English . they use machine translation, word alignment, sense projection, and sense filtering to produce silver annotations .
Outcome:	The proposed method produces parallel sense-annotated datasets on Farsi, Chinese, and Bengali . the results are higher than those obtained with recent multilingual systems, the authors say .

Copied to clipboard

Ishaan Watts, Varun Gumma, Aditya Yadavalli, Vivek Seshadri, Manohar Swaminathan, Sunayana Sitaram

Deciphering Hate: Identifying Hateful Memes and Their Targets (2024.acl-long)

Challenge:	Evaluation of multilingual Large Language Models is challenging due to a variety of factors including the lack of benchmarks with sufficient linguistic diversity, contamination of popular benchmarks into LLM pre-training data and lack of local, cultural nuances in translated benchmarks.
Approach:	They evaluate 30 models across 10 Indic languages by conducting 90K human evaluations and 30K LLM-based evaluations.
Outcome:	The proposed models perform best in most Indic languages, while the agreement drops for direct assessment especially for Bengali and Odia.

Copied to clipboard

Eftekhar Hossain, Omar Sharif, Mohammed Moshiul Hoque, Sarah Masud Preum

A Multilingual Parallel Corpora Collection Effort for Indian Languages (2020.lrec-1)

Challenge:	a growing body of research has focused on the negative aspects of memes in high-resource languages like Bengali . a new dataset for Bengali hateful memes is designed to detect their targeted entities .
Approach:	They propose a multimodal dataset that analyzes the modality of memes and compares them with other datasets.
Outcome:	The proposed dataset outperforms state-of-the-art datasets on Bengali hateful memes . the proposed dataset is generalizable on other low-resource hateful memes datasets compared with baselines based on the proposed model .

Copied to clipboard

Shashank Siripragada, Jerin Philip, Vinay P. Namboodiri, C V Jawahar

Multilingual Offensive Language Identification with Cross-lingual Embeddings (2020.emnlp-main)

Challenge:	Currently, neural network based approaches for machine translation are data hungry and sentence-level aligned parallel pairs are the currency.
Approach:	They propose to build sentence aligned parallel corpora across 10 Indian languages using online sources which have content shared across languages.
Outcome:	The proposed corpora significantly extends existing resources that are either not large enough or are restricted to a specific domain (such as health).

Copied to clipboard

Tharindu Ranasinghe, Marcos Zampieri

Same Claim, Different Judgment: Benchmarking Scenario-Induced Bias in Multilingual Financial Misinformation Detection (2026.findings-acl)

Challenge:	Several studies investigating methods to detect offensive content in social media use English data.
Approach:	They apply cross-lingual contextual embeddings and transfer learning to make predictions in languages with less resources.
Outcome:	The proposed method compares favorably to the best systems submitted to recent shared tasks on Bengali, Hindi, and Spanish.

Copied to clipboard

Zhiwei Liu, Yupeng Cao, Yuechen Jiang, Mohsinul Kabir, Polydoros Giannouris, Chen Xu, Ziyang Xu, Tianlei Zhu, Md. Tariquzzaman, Triantafillos Papadopoulos, Yan Wang, Lingfei Qian, Xueqing Peng, Zhuohan Xie, Ye Yuan, Saeed Almheiri, Abdulrazzaq Alnajjar, Ming-Bin Chen, Harry Stuart, Paul Thompson, Prayag Tiwari, Alejandro Lopez-Lira, Xue Liu, Jimin Huang, Sophia Ananiadou

BnMMLU: Measuring Massive Multitask Language Understanding in Bengali (2026.findings-acl)

Challenge:	Existing research on LLM biases has focused on direct questioning or general-purpose settings . pronounced behavioral biase despite their growing deployment in financial analysis, forecasting, and decision support.
Approach:	They propose a benchmark to evaluate behavioral biases of large language models in MFMD . they use a multilingual financial misinformation dataset to integrate these with misinformation claims .
Outcome:	The proposed benchmark evaluates behavioral biases of large language models across economic scenarios.

Copied to clipboard

Saman Sarker Joy, Swakkhar Shatabda

SHONGLAP: A Large Bengali Open-Domain Dialogue Corpus (2022.lrec-1)

Challenge:	Large-scale multitask benchmarks have driven rapid progress in language modeling, yet most emphasize low-resource languages like English.
Approach:	They propose a benchmark for massive multitask language understanding in Bengali . they use a dataset that preserves mathematical content via MathML and a subset of questions most frequently missed by top systems to stress difficult cases.
Outcome:	The proposed benchmark covers 24 model variants across 11 LLM families.

Copied to clipboard

Syed Mostofa Monsur, Sakib Chowdhury, Md Shahrar Fatemi, Shafayat Ahmed

Dialogue Systems & Conversational Agents

Challenge:	Existing open-domain dialogue systems suffer from data scarcity due to unavailability of high-quality datasets for low-resource languages like Bengali.
Approach:	They propose to prepare large-scale open-domain dialogue datasets from podcasts and talk-shows and label them based on weak-supervision techniques.
Outcome:	The proposed corpus improves performance of large language models in case of downstream classification tasks during fine-tuning.

Role of Language Relatedness in Multilingual Fine-tuning of Language Models: A Case Study in Indo-Aryan Languages (2021.emnlp-main)

Copied to clipboard

Tejas Dhamecha, Rudra Murthy, Samarth Bharadwaj, Karthik Sankaranarayanan, Pushpak Bhattacharyya

Improving Bengali and Hindi Large Language Models (2024.lrec-main)

Challenge:	A study of multilingual fine-tuning yields better performance on downstream NLP applications . low resource languages such as Oriya and Punjabi are found to be the largest beneficiaries of multi-lingual fine tuning.
Approach:	They propose to leverage the relatedness of languages that belong to the same family in NLP models by multilingual fine-tuning.
Outcome:	The proposed approach improves performance on downstream NLP tasks by 15% compared to monolingual fine-tuning.

Copied to clipboard

Arif Shahriar, Denilson Barbosa

IndicFinNLP: Financial Natural Language Processing for Indian Languages (2024.lrec-main)

Challenge:	Bengali and Hindi are low-resource languages, and the state-of-the-art tokenization methods fail to separate roots from affixes.
Approach:	They used BERT and Wordpiece tokenizers to train a wordpiece tokenization system for Bengali and Hindi to model fine-grained character-level information.
Outcome:	The proposed tokenizers outperform the state-of-the-art and Wordpiece tokenizer for modeling Bengali and Hindi.

Copied to clipboard

Sohom Ghosh, Arnab Maji, Aswartha Narayana, Sudip Kumar Naskar

Challenge:	IndicFinNLP is a collection of 9 datasets relating to FinNLP for three Indian languages.
Approach:	They propose to use financial NLP to detect exaggerated numerals in financial texts written in Hindi, Bengali, and Telugu.
Outcome:	The proposed framework detects exaggerated numerals in financial texts written in Hindi, Bengali, and Telugu.

IndicSpeech: Text-to-Speech Corpus for Indian Languages (2020.lrec-1)

Copied to clipboard

Nimisha Srivastava, Rudrabha Mukhopadhyay, Prajwal K R, C V Jawahar

BenNumEval: A Benchmark to Assess LLMs’ Numerical Reasoning Capabilities in Bengali (2025.findings-acl)

Challenge:	India has 22 languages, each of them being spoken by over a million people . the current state of the art text-to-speech systems for Indian languages are lacking in the multimedia domain .
Approach:	They propose to train a state-of-the-art TTS system for Hindi, Malayalam and Bengali and publish the results.
Outcome:	The proposed system trains neural text-to-speech systems for Hindi, Malayalam and Bengali and makes them publicly available.

Copied to clipboard

Kawsar Ahmed, Md Osama, Omar Sharif, Eftekhar Hossain, Mohammed Moshiul Hoque

Low-Resource NLP

Challenge:	Large Language Models (LLMs) excel in general-purpose tasks but struggle with numerical reasoning, especially in low-resource languages like Bengali.
Approach:	They propose a benchmark to assess LLMs on numerical reasoning tasks in Bengali.
Outcome:	The proposed benchmark assesses LLMs on numerical reasoning tasks in Bengali.

DIMSIM: Distilled Multilingual Critics for Indic Text Simplification (2024.findings-acl)

Copied to clipboard

Sneha Mondal, Ritika Ritika, Ashish Agrawal, Preethi Jyothi, Aravindan Raghuveer

BanglaAbuseMeme: A Dataset for Bengali Abusive Meme Classification (2023.emnlp-main)

Challenge:	Existing approaches to improve the quality of responses generated by large language models (LLMs) however, these critique-refine steps require multiple expensive LLM calls.
Approach:	They propose to use critique distillation to train critic models that are trained on input-critique pairs generated by an LLM.
Outcome:	The proposed model trains two separate critics that focus on lexical and structure complexity, and is more effective than using an LLM directly as a critic in both 0-shot and few-shot settings.

Copied to clipboard

Mithun Das, Animesh Mukherjee

A Culturally-diverse Multilingual Multimodal Video Benchmark & Model (2025.emnlp-main)

Challenge:	a number of studies have tried to detect and control the spread of such abusive memes on social media platforms.
Approach:	They build a Bengali meme dataset to test models for abusive memes . they find that multimodal models that use both textual and visual information outperform unimodal models .
Outcome:	The proposed model outperforms unimodal models in a Bengali meme dataset.

Copied to clipboard

Bhuiyan Sanjid Shafique, Ashmal Vayani, Muhammad Maaz, Hanoona Abdul Rasheed, Dinura Dissanayake, Mohammed Irfan Kurpath, Yahya Hmaiti, Go Inoue, Jean Lahoud, Md. Safirur Rashid, Shadid Intisar Quasem, Maheen Fatima, Franco Vidal, Mykola Maslych, Ketan Pravin More, Sanoojan Baliah, Hasindri Watawana, Yuhao Li, Fabian Farestam, Leon Schaller, Roman Tymtsiv, Simon Weber, Hisham Cholakkal, Ivan Laptev, Shin’ichi Satoh, Michael Felsberg, Mubarak Shah, Salman Khan, Fahad Shahbaz Khan

IndiGEC: Multilingual Grammar Error Correction for Low-Resource Indian Languages (2025.emnlp-main)

Challenge:	Large multimodal models have gained attention for their effectiveness to understand and generate descriptions of visual content.
Approach:	They propose a multilingual Video LMM benchmark to evaluate video LMMs across 14 languages . they also introduce a machine translated multilingual video training set .
Outcome:	The proposed video LMM benchmark is designed to evaluate video Lmms across 14 languages including Arabic, Bengali, Chinese, English, French, German, Hindi, Japanese, Russian, Sinhala, Spanish, Swedish, Tamil, and Urdu.

Copied to clipboard

Ujjwal Sharma, Pushpak Bhattacharyya

FIND: Toward Multimodal Financial Reasoning and Question Answering for Indic Languages (2026.findings-acl)

Challenge:	despite growing interest in GEC, most research has focused on English due to the lack of benchmark datasets for low-resource lan-guages.
Approach:	They propose a new approach to generate high-quality synthetic data for GEC using monolingual corpora.
Outcome:	The proposed framework outperforms other monolingual methods in English, Hindi, Bengali, Marathi, and Tamil.

Copied to clipboard

Sarmistha Das, Vaibhav Vishal, Syed Ibrahim Ahmad, Sriparna Saha, Manish Gupta

Reasoning

Challenge:	Existing benchmarks for numerical reasoning in multilingual Indic languages are inadequate . e.g., FinVQA is a framework for evaluating financial numerical reasoning .
Approach:	They propose a framework that combines supervised fine-tuning with constraint-aware decoding to promote faithful numerical reasoning.
Outcome:	The proposed framework spans English, Hindi, Bengali, Marathi, Gujarati, and Tamil . it combines supervised fine-tuning with constraint-aware decoding to promote faithful numerical reasoning .

When Meaning Travels: A Granular Lens on Hybrid-MoE’s Role in Idiomatic Understanding for Language Models (2026.findings-acl)

Copied to clipboard

Sarmistha Das, Vaibhav Vishal, Shreyas Guha, Amaan Ali, Kitsuchart Pasupa, Sriparna Saha

GanitLLM: Difficulty-Aware Bengali Mathematical Reasoning through Curriculum-GRPO (2026.findings-acl)

Challenge:	idioms provide a fascinating gateway to creativity, cultural values, historical context, and diverse perspectives inherent to diverse linguistic traditions.
Approach:	They propose a multimodal idiom corpus enriched with seven idiomatic tones . they propose idiomic hybridization framework that embeds multiple idiomatic expert opinions .
Outcome:	The proposed framework achieves 5–6% performance gains across advanced vision language models.

Copied to clipboard

Shubhashis Roy Dipta, Khairul Mahbub, Nadia Najjar

Reasoning

Challenge:	Existing LLMs either reason in English and translate, or simply fail on multi-step Bengali math.
Approach:	They propose a Bengali mathematical reasoning model called GanitLLM with a difficulty-aware Bengali math corpus and a curriculum-based GRPO pipeline.
Outcome:	The proposed model improves on Bn-MGSM and Bn MSVAMP by +8 and +7 accuracy points while increasing the percentage of Bengali reasoning tokens from 14% to over 88% and reducing solution length from 943 to 193 words.

Mind the Pause: Disfluency-Aware Objective Tuning for Multilingual Speech Correction with LLMs (2026.acl-long)

Copied to clipboard

Deepak Kumar, Baban Gain, Asif Ekbal