ARSENAL-GSD: یک چارچوب برای برآورد اعتماد در تیم های مجازی بر اساس تحلیل احساسات ARSENAL-GSD: A framework for trust estimation in virtual teams based on sentiment analysis

1. Introduction Software development using virtual teams characterizes distributed software development or global software development (GSD) when the distance between members comprises continents. It aims at providing benefits such as: low costs, proximity to the market, innovation and, access to skilled labor [1]. However, geographic distribution and cultural differences bring some challenges as well, mainly in communication, which depends mostly on computer mediated communication (CMC). One of the challenges faced by virtual teams and therefore GSD is to generate and sustain trust among team members. There are several studies that show the importance of trust for GSD teams [2–6], the benefits of having high trust teams, and the drawbacks of the lack of trust among members. Trust between developers that are in different and distant locations facilitates collaboration [7], effective knowledge sharing, conflict resolution and teams integration [8]. Thus, trust impacts the teams efficiency, since high trust teams can achieve their goals with less effort than low trust teams. Conversely, the lack of trust brings additional challenges to the team. As reported by Calefato et al.[7], a low level of trust aggravates the feeling of being separated by different objectives, reduce the willingness to share information and cooperate for problem solving, and affects the goodwill with others. So, in this context, information about trust among people can be used to suggest members to a team and/or to monitor the relationship among members, for example. Some models have been proposed to estimate trust among people based on trust evidences, such as number of interactions, success of these interactions and similarity among people [9,10]. We consider trust evidence something that indicates the existence of trust or that happens when there is trust among people. The information used by trust models can be extracted, for example, from social networks. In general, this information refers to the amount of interactions, evaluation of these interactions and their success. However, in a working environment people may not feel free to provide assessments of co-workers. Besides that, when the number of interactions is high, people may start to provide incorrect ratings, leading to incorrect trust estimation. Skopik et al. [9] developed a set of metrics to analyze the success of an interaction based on the fact that the bigger the amount of successful interactions the greater is the trust among people. These metrics eliminate the need for feedback, however, they are domain dependent and ignore subjectivity, one of the characteristics of trust, since they treat people as everyone thinking in the same way, which is not true. In this paper we present a framework to estimate trust among members of GSD teams called ARSENAL-GSD. It extracts trust evidences observed in member interactions on versioning systems, without human intervention and using sentiment analysis. In Sections 2–4 the concepts of GSD, trust and sentiment analysis, used in the development of the proposed framework are presented. Section 5 describes the framework ARSENAL-GSD. In Section 6 we evaluated the trust evidences and formulas used in the framework. Finally, Section 7 presents the conclusion and directions for future works.