Mohsinul Kabir

The rapid advancement of pre-trained language models has significantly enhanced performance across various natural language processing (NLP) tasks through task-specific fine-tuning. However, these models often require large, domain-specific annotated datasets for fine-tuning, which poses a significant challenge in low-resource domains such as under-represented culture, mental health, biomedical text, and literary analysis. The scarcity of annotated data in these areas limits the applicability of fine-tuning approaches, hindering progress in NLP research for resource-constrained domains. This research project aims to explore the potential of large language models (LLMs) as a solution to the lack of annotated datasets in low-resource linguistic domains. By leveraging zero-shot and few-shot prompt techniques, LLMs can perform various tasks without the need for extensive domain-specific fine-tuning. The study will investigate the generalization capabilities of LLMs, particularly in low-resource settings, to assess their effectiveness in addressing challenges inherent in these domains. The findings are expected to contribute to the development of more inclusive and accessible NLP tools that can operate efficiently even in the absence of large annotated corpora, thereby advancing the state of NLP research in low-resource environments.