Md. Masudur
Rahman

I'm an Assistant Professor in the Department of Computer Science and Engineering at the University of Global Village in Barishal, where I have taught since 2022. I completed my M.Sc. in CSE at Khulna University of Engineering & Technology in 2026 under Prof. Dr. K. M. Azharul Hasan.

My recent work builds BanEmoHate, a multilevel Bangla dataset of 16,407 statements drawn from natural social media comments, annotated for hate speech and emotion at the same time and validated with Cohen’s and Fleiss’ kappa. The premise is that the two signals inform each other: text carrying hate almost always carries an identifiable emotion, and that emotion turns out to be a useful feature for detecting the hate. The dataset is openly available on Mendeley Data. Earlier work covered emotion recognition from microblogs, particle swarm optimization for discrete problems, and applied machine learning for medical prediction and plant disease classification.

I supervise undergraduate theses and welcome collaboration on low-resource NLP, applied computer vision, and metaheuristic optimization.

Annotating Bangla text for hate speech alone throws information away. A sentence is angry, or sad, or cheerful, and that emotional register is part of what makes it hateful in the first place. BanEmoHate labels both levels on the same text: first hate or non-hate, then one of three emotions inside each branch. Six combined classes, one pass of annotation, and a model that can learn either level from the other.

Because the two levels sit on the same sentence, a classifier trained on one can be evaluated against the other. That is what the emotion-aided detection paper tests: whether knowing a comment's emotional register measurably improves the hate decision, rather than treating the two as unrelated tasks that happen to share a corpus.