Computer-Assisted Qualitative Visual Analysis: Automating Thematic Analysis of Images
Date
Authors
Supervisor
Item type
Journal Article
Degree name
Journal Title
Journal ISSN
Volume Title
Publisher
Intellect
Abstract
The advent of advanced artificial intelligence (AI) and machine learning technologies has opened new avenues for qualitative research, particularly in visual data analysis. This pilot study introduced computer-assisted qualitative visual analysis (CQVA), leveraging GPT-4 Turbo and Google Cloud Vision to automate the thematic analysis of visual datasets. Traditional methods, relying on manual coding, are time-consuming and labour-intensive. CQVA addresses these challenges by providing an efficient, scalable and cost-effective alternative. This study had two objectives: developing the CQVA method and applying it to analyse the top 1000 advertisements from the ‘adPorn’ subreddit, offering insights into Reddit users’ advertising preferences. A clear preference was identified for ads utilizing visual metaphors, as these were the most common. Additionally, the importance of engaging visual communication was underscored, with themes employing visually striking and easily comprehensible imagery being favoured by Reddit users. Despite its promise, CQVA required human intervention to guide AI outputs and validate clusters and themes. However, the findings demonstrated CQVA’s potential to revolutionize qualitative visual analysis by significantly reducing time and cost, while maintaining the richness of insights typically achieved through manual methods, thus enabling more efficient and comprehensive analysis of large visual datasets, highlighting the method’s scalability and practicality for future research.Description
Keywords
Source
Interactions: studies in communication and culture, ISSN: 1757-2681 (Print); 1757-2681 (Online), Intellect, 13(2), 147-167. doi: 10.1386/iscc_00058_1
Publisher's version
Rights statement
Contributors to all Intellect products can deposit their author accepted manuscript (AAM) in a non-commercial institutional or subject repository. We define an AAM as the version of the paper after peer review, with revisions having been made, but before copy-editing and typesetting have taken place. This is subject to an embargo period of twelve months.