Inferring success of online development communities: Application of graph entropy for quantifying leaders' involvement

In recent years, social activities have been rapidly shifting into networked environment, and therefore several types of online communities have been formed. For facilitating online communities and leading those online communities to success, it is essential for community administrators to adequately know the current state of online communities. In the literature, there exist several researches on factors affecting online community success. These studies suggest that leaders' involvement is one of major success factors in online communities. On the contrary, there exist researches on social networks, which are graphical representations of social relationships among people. For instance a technique for discovering important nodes based on graph entropy, which is obtained from a social network, is proposed. In this paper, we propose an index called LSI (Leadership Strength Index), which is an estimate of leaders' involvement in an online community obtained from its social network. We investigate properties of the proposed LSI through experiments utilizing log data of activities in SourceForge, which is one of major forums for online development communities. Consequently, we exhibit the potential of the proposed LSI for inferring success of online development communities.