A Machine-Based Personality Oriented Team Recommender for Software Development Organizations

Abstract

Hiring the right person for the right job is always a challenging task in software development landscapes. To bridge this gap, software firms start using psychometric instruments for investigating the personality types of software practitioners. In our previous research, we have developed an MBTI-like instrument to reveal the personality types of software practitioners. This study aims to develop a personality-based team recommender mechanism to improve the effectiveness of software teams. The mechanism is based on predicting the possible patterns of teams using a machine-based classifier. The classifier is trained with empirical data (e.g. personality types, job roles), which was collected from 52 software practitioners working on five different software teams. 12 software practitioners were selected for the testing process who were recommended by the classifier to work for these teams. The preliminary results suggest that a personality-based team recommender system may provide an effective approach as compared with ad-hoc methods of team formation in software development organizations. Ultimately, the overall performance of the proposed classifier was 83.3%. These findings seem acceptable especially for tasks of suggestion where individuals might be able to fit in more than one team.