Employing Structural Topic Modeling to Deconstruct Discourses and Trends Relating to Violence Against Women in Hong Kong
摘要
Despite the discourses relating to violence against women (VAW) in Hong Kong being conceptualized, limited empirical research covers the emerging text in online media and the current trends and dynamics of these developing discourses. This study aims to explore the online discourses relating to VAW in Hong Kong while deconstructing the prevailing topics, trends, and the interrelationship between these topics.
MethodsWe employed structural topic modeling (STM) to examine the discourses relating to VAW. The effect of temporal factors and related influence on discursive constructions between data sources were revealed by incorporating metadata. The analysis includes 2,562 online posts and comments between 2016 and 2024 collected from two media websites, including HK01 and LIHKG.
ResultsWe found 11 prevailing topics leading to the social constructions of VAW in Hong Kong. "Leave and Record," "Labeling", and "Female Subordination" have high prevalence across all topics. There are 5 groups of topic networks forming a dominant picture of VAW discourses in Hong Kong. The patterns of how these two websites constructed the discourses of VAW were identified.
ConclusionsOur findings expand the existing conceptualization of discourses relating to VAW within the Hong Kong context and enrich the understanding of dynamic trends and patterns prevailing in the Chinese community. The multiple structured masculinities might serve as the roots of these discursive patterns and have considerably reinforced female subordination. Additionally, this study highlights the advantages of machine learning techniques and advances the understanding of employing STM in researching human expressions of meaning.