Gender classification of product reviewers in China: a data-driven approach

Online product discussion forums have become essential resources for marketers seeking to understand market dynamics and consumer preferences. Identifying the gender of forum participants can further enhance the effectiveness and efficiency of marketing efforts. However, the relationship between linguistic features and gender classification often varies due to contextual factors such as genres, social networks, and social classes. Recognizing that the discriminatory power of gender markers changes with context, this study proposes and validates a framework to guide the adoption of existing gender classification systems specifically for online product discussions. We demonstrate that beyond optimizing the classification methods themselves, performance can be improved by strategically applying these methods to archived discussion data. Our findings reveal that, for a given classification method and discussion forum, the size of the input data significantly influences performance, with an optimal data size existing to achieve the best results.