Author Profiling on Social Media using New Weighting Schemes that Emphasize Personal Information

This paper summarizes the thesis: Identificacion del perfil de autores en redes sociales usando nuevos esquemas de pesado que enfatizan informacion de tipo personal" whose main idea indicates that terms located in phrases exposing personal information are highly valuable for the AP task. Firstly, it is presented an study on the relevance of this information to this task. Secondly, it is proposed a novel approach, which aims to emphasize the value of this type of terms by two proposals: a feature selection method and a term weighting scheme; both of them are based on a novel measure called personal expression intensity, which estimates the quantity of personal information revealed by each term. The approach was evaluated in age and gender prediction on different social media. The results are encouraging, with average improvements about 7.34% and 5.76% for age and gender identification respectively in comparison with the best results from the state of the art. These results allow concluding that personal information play an important role in the task.