Creation of a Corpus in Spanish for the Recognition of Personality Traits

Abstract

Automatic personality recognition is an area of research that has become very important in recent years. Currently there is research with different approaches that seek to automatically recognize personality traits by means of text. There are methods that use texts from voice transcriptions or texts written by people to determine whether an individual has a certain personality trait. These methods are based on machine learning and deep learning algorithms. A key element for the construction of such models is to have a data set (corpus) for training and optimization. This paper presents the creation of a corpus called PersonText, which contains 213 texts in Spanish with their respective labels related to the presence or absence of the personality traits of the Big-Five model, as well as the scores obtained by the participants in a standardized personality test. The main motivation for the creation of this corpus was the limited existence of corpora of texts in Spanish focused on personality recognition. The information was obtained from a platform developed by the research team, used for data collection based on standardized personality tests and videos of the participants. Additionally, to evaluate the corpus, tests were performed with different machine learning and deep learning models. The results obtained are promising and validate the relevance of the corpus built to address the task of automatic personality recognition.