Abstract
The concept of affinity relationship discovery is relatively new in the context of online discussion communities and there has been little work addressing it to date. This problem entails finding affinity relationships in a community by combining structural features and the content of interactions. Affinity discovery seeks not only to identify these affinity relationships, but also to quantify them so that the degree of affinity between individuals can be perceived in the form of a score. This paper proposes an algorithm based on Markov chain models, named HAR-search, for discovering hidden affinity relationships and deriving affinity scores between individuals in an online community. We demonstrate that our method is capable of tracking the evolution of affinity over time and predicting affinity relationships arising from the influence of certain community members. Comparison with state-of-the-art methods shows that our method results in robust discovery and considers minute details.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have