Uma abordagem computacional para identificação de indício de preconceito em textos baseada em análise de sentimentos

Abstract

There is a growing interest in detecting abusive language, hate speech, cyber bullying in recent years. Web sites and social networks have also been under increasing pressure to address these issues. Hate speech is usually defined as any communication that disparages a person or group based on some characteristics, such as race, color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristic. In the area of Artificial Intelligence, text mining can be defined as a set of techniques and processes for discovering innovative knowledge from textual data. Among the text mining techniques is Sentiment Analysis, or also known as opinion mining, act with the study of opinions, feelings, evaluations, attitudes and emotions of people in relation to entities such as products, services, organizations, individuals, problems. This paper proposes a possible solution for the discovery of evidence of prejudice in texts in Brazilian Portuguese, where a hybrid approach was developed combining approaches based on machine learning and lexical dictionaries. In addition, the approach was used in a pilot study to identify biased comments in essays.