Caching for Pattern Matching Queries in Time Evolving Graphs: Challenges and Approaches
Muhammad Nisar, Sahar Voghoei and Lakshmish Ramaswamy
University of Georgia, University of Georgia, University of Georgia

Pattern matching is an important class of problems related to graphs. It is a fundamental problem for many applications and has been extensively studied in literature. With the advent of huge graphs, the challenges in this domain have increased manifold. Consequently a lot of recent research has led to new architectures and approaches for optimized solutions to the pattern matching problem. A vast majority of these graphs hardly remain static and constantly evolve over time (like social networks, web graphs, etc). Recently, caching has been studied in the context of static graphs to optimize the throughput of query processing systems. In this paper, we list the challenges in caching in the context of Time Evolving Graphs (TEGs). Amongst others, one major challenge is consistency which entails to making sure the cache is consistent with the streaming changes. We propose an approach to successfully implement caching that addresses those issues and based on the initial results, we see significant gains in the overall performance of system.