From laborious to lucid : best practice for using generative AI for thematic and sentiment analysis

Kocheva, Nataliya and Fitkov-Norris, Elena (2024) From laborious to lucid : best practice for using generative AI for thematic and sentiment analysis. In: FBSS Research Conference; 21 Jun 2024, Kingston upon Thames, U.K.. (Unpublished)

Abstract

Thematic analysis, a cornerstone of qualitative research, often entails laborious procedures. Recent strides in Generative AI, such as ChatGPT, offer promising avenues for refining qualitative data analysis. This study probes the effectiveness of ChatGPT as a tool for enhancing thematic and sentiment analysis. Through 4 semi-structured interviews and 200 customer reviews from Trustpilot, we unearthed opportunities, challenges and reservations regarding ChatGPT's integration into qualitative analysis in social science. We suggest best practices for prompt engineering to augment generative AI’s role in thematic analysis. Results indicate that generative AI closely matches human-generated themes, providing an effective tool, capable of supporting thematic and sentiment analysis in social science research.

Item Type:	Conference or Workshop Item (Paper)
Event Title:	FBSS Research Conference
Organising Body:	Kingston University
Research Area:	Research Areas > Business and management studies Research Areas > Computer science and informatics Domains Case Studies > Business Domains Case Studies > Pedagogy Domains Case Studies > Teaching & Learning Research Areas > Education
Faculty, School or Research Centre:	Kingston Business School > Department of Accountancy, Finance and Informatics
Date Deposited:	22 Apr 2025 15:24
Last Modified:	22 Apr 2025 15:24
URI:	https://eprints.kingston.ac.uk/id/eprint/56946

Actions (Repository Editors)

Item Control Page