This paper investigates gender differences in online social media through a feature-based text classification framework. It analyzes writing styles and topics of interest to distinguish between male and female posters in a community-based Islamic women's political forum. The study demonstrates that incorporating both content-free and content-specific features into the classification framework leads to better performance compared to using only content-free features. Additionally, feature selection significantly enhances classification results. The findings reveal that female and male participants have notably different topics of interest, highlighting the utility of the proposed framework in understanding gender differences in online discussions.