A hybrid recommender system for finding relevant users in open source forums

Abstract

Open source projects rely heavily on online forums as a key input to the requirements process. These forums are valuable sources for information about the users and their needs. Part of the success of open source projects depends on the collaboration and synergy of community members as they engage in active and productive discussions through posting comments, questions, and advice to online forums. However, the lack of feedback which occurs when initial posts go unanswered can negatively affect the users' perception of the project, and can subsequently impede adoption, create frustration, and lead to loss of opportunities from not understanding and satisfying the users' needs. This problem is quite common in open source forums. Our recent analysis of seven open source projects found that anywhere from 14% to 37% of user posts never get a reply. This paper directly addresses the problem of unanswered posts by presenting a hybrid recommender system that can be used to identify potential users who might be capable of responding to unanswered posts. The proposed system was evaluated using a statistical cross validation, and results show that it significantly outperformed a benchmark random recommender in terms of precision and recall. In addition, an informal analysis of the relationships between the users and the threads is presented to provide further evidence for the potential of recommender systems in this area.