Determinación del género de autores de textos cortos a través de n-gramas

Nowadays, the possibilities for communicating or expressing through an electronic way are very wide: e-mail, social networks, chats and other ways are used by the majority of computer and mobile device users. One of the problemas that is presented in this communication way is excess, such as plagiarism, identitiy falsification, blackmailing, etc. Text authorship attribution (TAA) is in charge of answering authoring issues by providing previous examples from said author (training set). A useful process within TAA is sex or gender identification (male, female), which has been studied by many authors for its use in English mostly. The present work proposes a computational model 57 Research in Computing Science 115 (2016) pp. 57–66; rec. 2016-04-22; acc. 2016-05-18 based on lexical characteristics (n-grams) for gender identification in short texts in Spanish. Tests were carried out with a corpus from social network and blog text messages, producing promising results.