A Compressed Suffix-Array Strategy for Temporal-Graph Indexing

Abstract

Temporal graphs represent vertexes and binary relations that change over time. In this paper we consider a temporal graph as a set of 4-tuples ( v s , v e , t s , t e ) indicating that an edge from a vertex v s to a vertex v e is active during the time interval [t s , t e ). Representing those tuples involves the challenge of not only saving space but also of efficient query processing. Queries of interest for these graphs are both direct and reverse neighbors constrained by a time instant or a time interval. We show how to adapt a Compressed Suffix Array (CSA) to represent temporal graphs. The proposed structure, called Temporal Graph CSA (TGCSA), was experimentally compared with a compact data structure based on compressed inverted lists, which can be considered as a fair baseline in the state of the art. Our experimental results are promising. TGCSA obtains a good space-time trade-off, owns wider expressive capabilities than other alternatives, obtains reasonable space usage, and it is efficient even when performing the most complex temporal queries.