Exploring Gender Differences in Fatwa through Machine Learning

Abstract

This paper focuses on exploring the differences in inquiries made by men and women within a religious context. Additionally, we aim to ascertain whether it’s feasible to forecast the popularity of answers and the factors contributing to their popularity. To achieve this, we compile a new dataset comprising 40,000 question-answer pairs categorized by gender and popularity. These are sourced from online question-and-answer platforms. Our methodology involves comprehensive experimental analysis, utilizing advanced Arabic text preprocessing alongside machine learning algorithms. We concentrate on two primary objectives: predicting the gender of the questioner and forecasting the popularity of answers. Furthermore, we delve into thematic variations based on gender and address pivotal research queries that offer new perspectives within this domain. These include investigating the differences between questions posed by women versus men, exploring the potential for automated classification of queries by gender, predicting the popularity of fatwas, and identifying the contributing factors to their popularity. Our experimental findings demonstrate a 98% accuracy in gender prediction, precise predictions of popularity with minimal margin for error, and the identification of topics and their associations that are more inclined towards either men or women. We intend to share both the dataset and the source code openly with the research community.