Applications of Optimal Stopping Algorithm for Social Graph Based Recommendation

Recommendation Systems are becoming one of the most used tools in online marketing. In this paper, basic item based recommendation algorithms were extended to make use of a user's social circle's influence in deciding the recommendations. Depending on the social network of a user, the items to be recommended were selected. Apart from this, his own previous ratings of movies were considered to get an assessment of his taste and this was added up to this ego-network's preferences to get a list of recommendations. An optimal stopping algorithm was then implemented on top of this recommendation system. This algorithm helped us pick users better from the ego-network. In the end, a general recursive optimal stopping algorithm is proposed, providing an efficient way to search for best set of users to recommend movies from throughout the social network.