Database Theory in Action: From Inexpressibility to Efficiency in GQL's Order-Constrained Paths

TL;DR

This paper introduces a translation technique for GQL's path queries that overcomes expressiveness limitations and improves practical query performance in graph databases like Neo4j.

Contribution

It presents a constructive translation that encodes increasing-edge constraints into the input graph, enhancing both expressiveness and efficiency.

Findings

The translation restores expressiveness for order-constrained path queries.

The compiled queries run faster and avoid timeouts in Neo4j's Cypher.

The approach demonstrates practical benefits of theoretical insights in graph query optimization.

Abstract

Pattern matching of core GQL, the new ISO standard for querying property graphs, cannot check whether edge values are increasing along a path, as established in recent work. We present a constructive translation that overcomes this limitation by compiling the increasing-edges condition into the input graph. Remarkably, the benefit of this construction goes beyond restoring expressiveness. In our proof-of-concept implementation in Neo4j's Cypher, where such path constraints are expressible but costly, our compiled version runs faster and avoids timeouts. This illustrates how a theoretically motivated translation can not only close an expressiveness gap but also bring practical performance gains.