Social Stream Data: Formalism, Properties and Queries

A social stream, which refers to the data stream that records a series of social stream entities and the dynamic relations between entities, and each entity created by one producer. It is not only can used to model user generate content in online social network services, but also a multitude of systems in which records are combined by graph and stream data. Thus, the research efforts in the area about social stream is one of the hot spots recently. Although the term of “social stream” have appeared frequently, we note there are rarely formal definitions and lacks a unified view on the data. In this paper, we formally define the social stream data model trying to explain the graph stream generating mechanism from the perspective of producers. Then several properties describing social stream data are introduced. Furthermore, we summarize a set of basic operators that are essential to analytic queries based on social stream data, describe their semantics in detail. A classification scheme based on query time window is provided and difficulties lies behind each type are discussed. Finally, three real life datasets are used for the experiment of calculating properties to reveal differences between different datasets and analyze how they may exacerbate hardness of queries.