Reputation Prediction Using Influence Conversion

Currently, due to constantly increasing popularity of social media sites and thus increasing amounts of data available to study, the field of social computing has gained momentum. The research about human interactions in social networks have proven to have multiple usages in e-commerce, recommendation and others. Amongst studied topics, notions of trustworthiness and influence drawn much attention in recent years. However, these notions were studied separately and independently, with little focus on the fact that, in real life, they tend to exist simultaneously. In this paper, we focus on this novel problem and present an investigation of both influence and reputation. In particular, we propose a transition method, that uses existing influence information from social network in order to predict the collective trustworthiness of the node, or reputation. Through preliminary experiments on a real-world dataset, we demonstrate the suitability of our method.