Multi-label semi-supervised classification applied to personality prediction in tweets

Lima A.C.E.S.; De Castro L.N.

Multi-label semi-supervised classification applied to personality prediction in tweets

item.page.type

Artigo de evento

Date

2013

item.page.ispartof

Proceedings - 1st BRICS Countries Congress on Computational Intelligence, BRICS-CCI 2013

item.page.citationsscopus

17

Authors

Lima A.C.E.S.
De Castro L.N.

Abstract

Social media allow web surfers to produce and share content about different subjects, exposing their activities, opinions, feelings and thoughts. In this context, online social media has attracted the interest of data analysis researchers seeking to infer behaviors and trends, besides creating statistics involving social sites. A possible research involving these data is known as personality analysis, which aims to understand the user's behavior in a social media. Thus, this paper uses machine learning techniques to predict personality traits in groups of tweets. In traditional approaches of personality prediction the analysis is made in the users' profiles and their tweets. However, in this paper the approach arises from the fact that the personality analysis is performed in groups of tweets. The prediction is based on the Big Five Model, also called Five Factor Model, which divides personality traits into five dimensions and uses linguistic information to identify these traits. This paper uses TV shows from Brazilian stations for benchmarking. The system achieved an average accuracy of 84%. © 2013 IEEE.

item.page.scopussubject

Big five , Multi-label classifications , Personality , Semi-surpevised learning , Twitter