A machine learning approach for gender identification of Greek tweet authors

Digital communities and social media are widely used and produce a huge amount of information every second. Text analysis has been widely used by researchers and machine learning (ML) engineers for automating the author profiling task. Author profiling can be used in marketing and business intelligence frameworks but also remains a strong factor in crime investigations gaining more insight regarding the suspect. In this paper we describe the process we used in obtaining a new Twitter corpus and propose an ML approach to determine the gender of Greek author's tweets. The best result (0.7 accuracy) was obtained using SVMs and TF-IDF encoding.