GraphDB 9.3 Speeds Up Graph Traversal

Optimized support for arbitrary path length in SPARQL brings quicker discovery of relationships in knowledge graphs 

Sofia/Bulgaria Tuesday, June 9, 2020

Ontotext has released the 9.3 version of GraphDB, which makes the most powerful RDF database engine substantially faster in graph traversal and analytics. This version also enhances the access to secured LDAP servers and updates GraphDB’s Connectors for Elasticsearch, Solr and Lucene to their latest versions.

The property paths represent one of the most powerful mechanisms in the SPARQL query language, which allows users to specify relationship patterns that the engine should apply to traverse multi-hop relationships between two nodes. For instance, the path ?company has-subsidiary / located-in Moscow will discover companies with subsidiaries in Moscow, using a sequence of two relationships: the first of which is of type has-subsidiary and the second of type located-in. Property paths can also include repetitive elements specified with the familiar notation for regular expressions: * (one or many), + (zero, one or many) and ? (zero or one) relationships of the corresponding type. This allows the definition of complex patterns like the following

?company has-subsidiary+ / registered-in / part-of+ / status Offshore

This example would retrieve companies, which, directly or via a chain of subsidiaries, are represented in a place that is part of a territory with offshore status. Property paths represent a compact and easy to comprehend syntax to express non-trivial patterns of relationships in graphs. The arbitrary path operators * and + let SPARQL queries that require the database engine to perform recursive traversal of the graph on multiple steps.

As of version 9.3, GraphDB handles property paths natively. This permits GraphDB to calculate better execution plans and answer queries with property paths much faster. For instance, some experimental queries with property paths against Wikidata now get evaluated in less than a second versus almost a minute before this optimization. Together with some performance optimizations in version 9.2.1, related to dealing with named graphs, this rounds up another important step to provide good and predictable query performance across all sorts of retrieval and analytic scenarios.

Version 9.3 also makes improvements and fixes related to enterprise security. Following the best practices for configuring LDAP servers, GraphDB now replaces the anonymous access with proper authentication. Lightweight Directory Access Protocol (LDAP) is widely used by enterprises to integrate directory services, which share information about users, systems, networks and applications throughout the network.

A complete list of changes and fixes are available in GraphDB 9.3 release notes.

Get the latest GraphDB now!

For more information, contact Doug Kimball, Chief Marketing Officer at Ontotext