Workload-aware cache for social media data

Abstract

The success of social network services has brought up many interesting Web 2.0 applications, while posed great challenges for human-real-time data management for huge volume of data with unstructured nature. Timeline query is a specific type of queries that are widely used in social network services and analysis. A workload-aware cache scheme for efficient evaluation of home timeline queries in human-real-time manner is proposed in this paper. It utilizes the communities within followship network and considers the high skew of access frequency across users to generate cache units. Thus, timeline queries are transformed to a process of merging cache units. Empirical studies show the superiority of overlapping cache strategy over other three existing strategies. © 2013 Springer-Verlag.