Published in SemEval @ ACL 2023, 2023
Knowledge base augmentation with pre-trained language models for multilingual Named Entity Recognition, achieving Top 10 rankings in 10 tracks at SemEval 2023 Task 2.
Recommended citation: Equal Contribution. (2023). "garNER at SemEval-2023 Task 2: Knowledge Augmentation for Multilingual NER." SemEval @ ACL 2023.
Published in BLP Workshop @ EMNLP 2023, 2023
An adaptive framework and annotated dataset for detecting social network posts that lead to communal violence, leveraging semi-supervised topic modeling with human-in-the-loop validation.
Recommended citation: Saha, S. et al. (2023). "Vio-Lens: Annotated Social Network Posts Leading to Communal Violence." BLP Workshop @ EMNLP 2023.
Published in BLP Workshop @ EMNLP 2023, 2023
The first shared task on violence-inciting text detection: task design, evaluation metrics, baselines, and error analysis of 27 participating teams and 16 accepted papers.
Recommended citation: Saha, S. et al. (2023). "BLP Shared Task 1: Violence-Inciting Text Detection." BLP Workshop @ EMNLP 2023.
Published in ACM FAccT 2024, 2024
A qualitative analysis of annotation practices in sensitive domains, introducing the FRS annotation method that addresses colonial and secular bias in data collection.
Recommended citation: Co-author. (2024). "Data, Annotation, and Meaning-Making: Faith-Based Communal Violence." ACM FAccT 2024.
Published in BUCC Workshop @ LREC-COLING 2024, 2024
A gold standard corpus for paraphrase detection in Bangla, featuring automated news headline collection, a carefully designed annotation schema, and benchmarking with language models and GPT.
Recommended citation: Saha, S. et al. (2024). "BnPC: A Gold Standard Corpus for Paraphrase Detection in Bangla." BUCC Workshop @ LREC-COLING 2024.
Published in Experimental & Molecular Medicine, Nature Publishing Group, 2025
A comprehensive survey and comparison of 40+ transcriptome-wide computational tools for alternative polyadenylation and splicing, covering long-read sequencing and ML-based isoform detection.
Recommended citation: Second Author. (2025). "Computational Methods for Alternative Polyadenylation and Splicing in Post-Transcriptional Gene Regulation." Experimental & Molecular Medicine. Nature Publishing Group, Q1, IF 12.8.
Published in Computational and Structural Biotechnology Journal (CSBJ), Elsevier, 2025
A comprehensive framework using genome language models for de novo polyadenylation site prediction, including dataset construction with hard negatives and genome-wide discovery experiments.
Recommended citation: Saha, S. et al. (2025). "PolyA-GLM: A comprehensive framework for De novo polyadenylation site prediction using genome language models." Computational and Structural Biotechnology Journal. Elsevier, Q1, Impact Factor 5.2.
Undergraduate courses, University of Asia Pacific (UAP), Department of CSE, 2024
Served as Lecturer at the University of Asia Pacific from February 2024 to July 2024. Taught the following undergraduate courses:
Undergraduate course, University of Central Florida, Department of Computer Science, 2024
Serving as Graduate Teaching Assistant for Introduction to Computer Concepts at the University of Central Florida since August 2024. Responsibilities include holding office hours, grading assignments, and supporting students in foundational computing topics.