Querying Hidden Attributes in an Online Community Network

An online community network such as Twitter, Yelp or amazon.com links entities (e.g., Users, products) with various relationships (e.g., Friendship, co-purchase, co-review) and make such information available for access through a web interface. Often, these community networks act as "social sensors" in which users sense information in the real world and mention them online. The web interfaces of these networks often support features such as keyword search that allow an user to quickly find entities of interest. While these interfaces are adequate for regular users, they are often too restrictive to answer complex queries such as (1) find 100 Twitter users from California with at least 100 followers who talked about earthquakes last year or (2) find 25 restaurants in Yelp with at least 10 5-star reviews with 10 or more 'useful' points. In this paper, we investigate the problem of answering complex queries that involve non-searchable attributes through the web interface of an online community network. We model such a network as a heterogeneous graph with two access channels, Content Search and Local Search. We propose a number of efficient algorithms that leverage properties of the heterogeneous graph and also propose a strategy selection algorithm based on the concept of multi-armed bandits. We conduct comprehensive experiments over popular social sensing websites such as Twitter and amazon.com which demonstrate the efficacy of our proposed algorithms.