Personalized social network activity feeds for increased interaction and content contribution

Abstract

Online social networks were originally conceived as means of sharing information and activities with friends, and their success has been one of the primary contributors of the tremendous growth of the Web. Social network activity feeds were devised as a means to aggregate recent actions of friends into a convenient list. But the volume of actions and content generated by social network users is overwhelming, such that keeping users up-to-date with friend activities is an ongoing challenge for social network providers. Personalization has been proposed as a solution to combat social network information overload and help users to identify the nuggets of relevant information in the incoming flood of network activities. In this paper, we propose and thoroughly evaluate a personalized model for predicting the relevance of the activity feed items, which informs the ranking of the feeds and facilitates personalization. Results of a live study show that the proposed feed personalization approach successfully identifies and promotes relevant feed items and boosts the uptake of the feeds. In addition, it increases the contribution of user-generated content to the social network and spurs interaction between users.