Interactive recommendations in social endorsement networks

An increasing number of social networking platforms are giving users the option to endorse entities that they find appealing, such as videos, photos, or even other users. We define this model as a Social Endorsement Network, visualized as a bipartite graph with edges (endorsements) from users to endorsed entities. In this work, we formalize the problem of interactive recommendations in social endorsement networks: given a query of tags and a social endorsement network, the problem is to recommend entities that match the query and also share a significant number of common endorsers. We propose an efficient search engine for the solution of the problem, able to produce high-quality and explainable recommendations. The entire framework is designed in a principled and efficient manner, making it ideal for large-scale systems. In a thorough experimental evaluation on real datasets, we illustrate the efficacy of our methods and provide some valuable insight on social endorsement networks. C...

• Type: Conference
• Year: 2010
• Where: RECSYS
• Authors: Theodoros Lappas, Dimitrios Gunopulos
• Keywords: Control Systems, Entities, RECSYS 2010, Social Endorsement Networks, Social Networking Platforms