Undergraduate Thesis Defended
I successfully defended my undergraduate thesis at the Department of Computer Science and Engineering, University of Dhaka.
The thesis, titled “Beyond Surface Metrics: Leveraging LLMs to Assess Semantic Depth of Human Responses,” was completed under the supervision of Dr. Md. Mosaddek Khan. It focuses on using large language models to move beyond shallow word-overlap metrics and evaluate the deeper semantic quality of student responses in educational settings.
This work was carried out as part of my B.Sc. journey and reflects my broader research interest in AI, natural language processing, educational technology, and human-aligned evaluation systems. The project explores how automated assessment systems can better align with human judgment while remaining practical for real educational use.
A central outcome of this work is CognifyQ, a web-based platform for semantic grading of written responses, designed to support both Bangla and English. The system aims to make assessment more consistent, scalable, and useful for both teachers and students.
I am deeply grateful to my supervisor, teachers, collaborators, and everyone who supported me throughout this journey. Defending this thesis marks an important milestone in my undergraduate life and a meaningful step toward my future research goals.
Thesis Details
- Thesis Title: Beyond Surface Metrics: Leveraging LLMs to Assess Semantic Depth of Human Responses
- Institution: University of Dhaka
- Department: Department of Computer Science and Engineering
- Supervisor: Dr. Md. Mosaddek Khan
- Project Focus: LLM-based semantic grading, educational AI, bilingual assessment, feedback quality
Looking Ahead
This thesis has strongly shaped my current research direction. Going forward, I hope to continue working on problems related to LLM evaluation, educational AI, multilingual NLP, and practical human-centered AI systems.
