Friends based keyword search over online social networks

Abstract

Online social networks are rapidly becoming popular for users to share, organize and locate interesting content. Users pay much attention to their close friends, those direct or two-hop friends. Users of Facebook commonly browse relevant profiles and the homepages, which are inefficient in obtaining desired information for a user due to the large amount of relevant data. In this paper, we propose a summary index with a ranking model by extending existing Bloom filter techniques, and achieve efficient full-text search over large scale OSNs to reduce inter-server communication cost and provide much shorter query latency. Furthermore, we conduct comprehensive simulations using traces from real world systems to evaluate our design. Results show that our scheme reduces the network traffic by 94.1% and reduces the query latency by 82.4% with high search accuracy. © 2013 Springer-Verlag.