Labeled graph sketches: Keeping up with real-time graph streams

Abstract

Currently, graphs serve as fundamental data structures for many applications, such as road networks, social and communication networks, and web requests. In many applications, graph edges stream in and users are only interested in the recent data. In data exploration, the storage and processing of such massive amounts of graph stream data has become a significant problem. As the categorical attributes of vertices and edges are often referred to as labels, we propose a labeled graph sketch that stores real-time graph structural information using only sublinear space and that supports graph queries of diverse types. This sketch also works for sliding-window queries. We conduct extensive experiments on real-world datasets in six different domains and compare the results with a state-of-the-art method to show the accuracy, efficiency, and practicability of our proposed approach.