Determinação da confiança no desenvolvimento global de software utilizando análise de sentimentos

Abstract

The advance in technology has enabled the emergence of a new work structure, virtual teams. In these teams, people are in different places and possibly over different time zone, making use of computer mediated communication. This structure provides great flexibility and some benefits such as to find skilled labor easier. However, distance creates some challenges, especially, regarding to means of communication. The means of communication in turn can bring greater difficulty in developing trust, essential for efficiency in these teams. In order to ensure greater e efficiency of these teams, we can use the trust established during previous projects in the process of members allocation to a new team and/or, during the project execution in order to monitor the relationship between members. Focusing on global software development teams, trust estimation can be performed using sentiment analysis of versioning systems comments, a kind of system used for software development, as well as other of its features. This master thesis presents an automatic framework, called ARSENAL-GSD, for trust estimation between members of global software development teams using sentiment analysis developed from trust evidences and trust models found in literature. The framework allows global software development team managers to evaluate interaction between members and their trust level. This information assists managers to select members to new teams and allows trust monitoring between members, ensuring more efficiency to the teams.