Análise De Dados Em Uma Rede Social Virtual Para Suporte Parental

Abstract

Virtual Social Networks have revolutionized the ways of social interaction, so much so that there is an expansion of its borders for applications in specific domains, such as academic production, product reputation, human resources management, among others. Due to the increasing engagement of users, such platforms daily generate a massive amount of unstructured data whose manual analysis is not feasible. In this scenario, the proposal of this work is to develop a computational environment capable of performing sentiment analysis and topic modeling in a social network of specific domain, in this case, the social network DEFICIÊNCIA, in order to identify trending topics, the predominant sentiment among users and the connection between topics and the network proposal. To accomplish this goal, it was necessary to analyze the recent literature of sentiment analysis methods adopted in online social networks analysis. Thus, a systematic literature review was carried out in the CAPES database considering three search queries which are formed from the combination of the search terms that contemplate the research scope. The searches results were evaluated through two relevance tests, which have an eliminatory nature. The first one was conducted by the authors and the second by two judges, these have a crucial role in the integrity of the review, would assess whether the data obtained are consistent with the focus of the study. From the results, it was possible to identify the methods of sentiment analysis prominent in studies involving social network analysis, identify the main research topics covered and the correlation between them. The network found presents a pattern compatible with a scale free network. Also, the network does not show a random pattern, but preferential attachment to specific keywords.