Exploration and Exploitation of Developers' Sentimental Variations in Software Engineering

Introduction

Emotions are inseparable part of human nature, which influence people’s activities and interactions, and thus emotions affect task quality, productivity, creativity, group rapport and job satisfaction (Choudhury & Counts, 2013; Feldt, Angelis, Torkara, & Samuelssonc, 2010; Palacios, LÓpez, Crespo, & Acosta, 2010). Software development, being highly dependent on human efforts and interactions, is more susceptible to emotions of the practitioners. Hence, a good understanding of the developers’ emotions and their influencing factors can be exploited for effective collaborations, task assignments (Dewan, 2015), and in devising measures to boost up job satisfaction, which, in turn, can result in increased productivity and projects’ success (Denning, 2012).

Several studies have been performed in the past for understanding the role of human aspects on software development and engineering. Some of those earlier studies address when and why employees get affected by emotions (Choudhury & Counts, 2013; Guzman, AzÓcar, & Li, 2014; Guzman & Bruegge, 2013; Pletea, Vasilescu, & Serebrenik, 2014; Tourani, Jiang, & Adams, 2014), whereas some other work address how (Graziotin, Wang, & Abrahamsson, 2013; Khan, Brinkman, & Hierons, 2010; Lesiuk, 2005; Murgia, Tourani, Adams, & Ortu, 2014; Wrobel, 2013) the emotions impact the employees’ performance at work. Despite those earlier attempts, software engineering practices still lack theories and methodologies for addressing human factors such as, emotions, moods and feelings (Graziotin, Wang, & Abrahamsson, 2015; Guzman & Bruegge, 2013). Hence, the community calls for research on the role of emotions in software engineering (Khan et al., 2010; Palacios et al., 2010; Shaw, 2004).

Some software companies try to capture the developers’ emotional attachments to their jobs by means of traditional approaches such as interviews and surveys (Wrobel, 2013). Capturing emotions with the traditional approaches is more challenging for projects relying on geographically distributed team settings and voluntary contributions (e.g., open-source projects) (Destefanis, Ortu, Counsell, Marchesi, & Tonelli, 2015; Guzman et al., 2014). Thus, to supplement or complement those traditional sources, software artifacts such as the developers’ commit comments/messages have been identified for the extraction of important information including developers’ emotional states (Guzman et al., 2014; Guzman & Bruegge, 2013; Pletea et al., 2014).