Exploring Personality Structure in South Africa: A Text Mining Approach
Date
2024-03-15
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of the Witwatersrand, Johannesburg
Abstract
Physical expression, behavioural attributes and social relations of an individual can often be studied through personality traits. This has made personality research a relevant aspect of gaining a deeper understanding of people in various contexts, for clinical reasons as well as social relatability. Trait theory has been fundamental in utilizing statistical methods such as factor analysis to construct the personality models that currently exist. The Five Factor Model (FFM) is amongst the most widely accepted of these trait theory models. Personality assessment instruments are developed as operationalisations of these models. These include the Goldberg Adjective Checklist, the South African Personality Inventory (SAPI), and the Chinese Personality Assessment Inventory (CPAI). Recently, naturally occurring data like social media statuses or Facebook Posts are being considered as data examining personality structure. This study aims to explore personality structure data obtained from South African literary texts and text mining techniques. Various techniques of text mining such as parts of speech tagging, and unsupervised and supervised LDA topic modelling were applied to 60 South African literary texts. While topic modelling showed limitations when used in an unsupervised manner, when guided by thematic clusters it presented comprehensible trait classifications that fit with the clusters as defined by the FFM. The instances where there was no fit corresponded with the literature which demonstrates poor fit for those constructs in African constructs. The results also showed that there is a difference in the expression of personality traits between men and women with the differences concurring with those found in the broader literature on gender differences across personality. While the text corpus for this study was small, there is evidence to suggest that text mining techniques could be used to assist in research on personality structure. Text mining is an approach that requires further research as it can be useful in dealing with large data that is naturally occurring to provide a better contextual exploration of personality.
Description
A research report submitted in partial fulfillment of the requirements for the degree of Master of Arts (E-Science) to the Faculty of Humanities, University of the Witwatersrand, Johannesburg, 2024
Keywords
Personality; Text Mining; Supervised Topic Modelling; Unsupervised Topic Modelling; SAPI; Latent Dirichlet Allocation (LDA); Five Factor Model
Citation
Gama, Beauty. (2024). Exploring Personality Structure in South Africa: A Text Mining Approach [Master’s dissertation, University of the Witwatersrand, Johannesburg].