An Ontology-based Approach for Analyzing Emotions in Software Developers’ Mailing Lists

  • Huma Tabassum
  • Sohaib Ahmed


Emotions are the feelings one has toward various entities. These have always intrigued scholars since early days. With the advancement of information and communication technologies, automated emotion analysis has been in the focus of research community since last decade. This is because people are extensively using online communications for different purposes such as social networking, writing blogs, tweets and product reviews. These online comments are growing enormously due to increasing number of users daily. Therefore, a concern raised by different communities including government, organizations and customers to analyze and explore these online comments for opinion mining and/or emotion analysis purposes. In software industry, software developers generally use forums, mailing lists and discussion groups in order to collaborate among themselves while developing software projects. However, developers may experience challenges such as less effective communications and conflicts during collaborative activities. Hence, this paper addresses to explore these challenges by identifying whether software developers may possess emotions while communicating in discussions. In the previous literature, ontology-based approach has been sparsely explored for emotion analysis, specifically for software developers’ collaborations. For this purpose, an ontology called EmotiOn, is used in order to perform emotion analysis on software developers’ mailing lists. Further, an analyzer is implemented that processes the design ontology. In this study, emails of two projects from archived Apache Software Foundation (ASF) mailing lists were used for analysis purposes. The results showed that 63% of those emails revealed various emotions. This indicates that software developers indeed express emotions through online communications. The analyzer was able to recognize correct emotions with a reasonable accuracy of 61.3%. The precision and recall measures for each of the emotions were recorded and presented in this paper.