A Multi-Modal Approach to Twitter User’s Gender Classification

Recent advancements in artificial intelligence (AI) and natural language processing (NLP) have revolutionized the identification of Twitter users by leveraging profile information and tweet content. AI and NLP technologies facilitate user identification, behavior analysis, and threat detection by scrutinizing profiles—such as usernames, biographies, and profile images—as well as tweet content. This paper explores the integration of profile and tweet data for user identification, focusing on the combination of profile information and tweet content to improve accuracy. The study also reviews gender and age classification research, noting remarkable improvements from classical methods to advanced deep learning approaches. The study classifies Twitter users' genders Using a multi-modal methodology that includes both image-based (VGG۱۶) and text-based (LaBSE) models. The combined model demonstrated superior performance compared to single-modal approaches, showcasing the effectiveness of feature fusion in enhancing classification precision. This multi-modal method improves valuable improvements for cybersecurity, marketing, and social analysis by providing more reliable user identification in the context of social media.

کلیدواژه ها:

Social network ، Author profiling ، Twitter ، Natural language processing (NLP) ، Deep neural networks ، Machine learning