Uma metodologia de geração automática de personas a partir de dados dos usuários em sistemas sociais

Abstract

Knowing the end user of a system is essential both for its development and for its evaluation. There is a user modeling technique called Personas, whose goal is to bring developers and users together, building people models, behaviors, and motivations for end-users. Some studies have proposed techniques to generate personas, collecting data from interviews and questionnaires. However, this type of generation may require a large manual analysis to increase accuracy, making the process exhaustive and time consuming. To solve such problems, computational resources can be used to automate the process of extraction and generation of personas. This study aims to present a proposal of automatic generation of Personas to assist in the process of system evaluation, using a textual evaluation methodology called MALTU. This work was executed in three steps: i) systematic mapping on automatic generation of personas, in order to understand the techniques; ii) definition of a proposal; and iii) experiment using data from the social system Twitter. With the accomplishment of the systematic mapping, no proposal of automatic generation of personas that has been applied in the context of systems evaluation was identified. With this, this study presents a proposal, which uses public data of users, together with the results obtained with the execution of textual evaluation. To complement the results of the mapping, a study of the twitter API was performed in order to understand what data can be extracted and how they can be analyzed. Finally, a proposal validation was performed, based on an experiment with Twitter posts. This experiment showed the validity of the proposal.