Towards a Strong Containment for Efficient Matching of Expressive Graph Patterns

We consider conditional graph patterns (CGPs) that allow querying data graphs with complex features like quantification, negation and predicates. To overcome the prohibitive cost of subgraph isomorphism, we consider matching of CGPs under simulation semantics which can be conducted in quadratic time. In emerging applications that deal with large data graphs, one would like to reduce more this matching time. We consider the optimization problem that checks whether match results of some pattern P1, over any data graph, can be extracted from those of another pattern P2. This allows an efficient matching of P1 using only match results of P2 which may be much smaller than the underlying data graph. We show that when patterns are very simple then the traditional containment can be used to meet our optimization goal. However, in case of complex pattern features, the containment semantics does not suffice since containment between two patterns does not imply a possible extraction between their match results. Hence, we propose an enhanced semantics, called strong containment, that consists in checking match results extraction between two patterns. We show that strong containment can be decided in cubic time for CGPs by providing such an algorithm. This new semantics can be applied in many emerging applications such as view-based answering, query optimization and caching systems maintenance.