Co-reference Resolution in Prompt Engineering

Abstract

Co-reference resolution is a longstanding research problem that researchers have been working to solve for the past fifty years. It is a crucial task in numerous Natural Language Processing (NLP) applications, such as Machine Translation (MT), Text Summarization, Question Answering, etc. Over time resolution strategies have evolved and incorporated new methods such as deep learning approaches. With the advancement of Large Language Models (LLMs), researchers are now tackling several problems using prompt engineering. This paper explores a pioneering attempt at Prompt Engineering in Co-reference resolution in the Assamese language. The Assamese language is resource-scarce, and this effort provides a way to overcome the resource barrier in the NLP domain. The experiment employs zero-prompt, few-prompt, and chain-of-thought prompt techniques, documenting the improvement in results and comparing them with the latest state-of-the-art in the Assamese language.