RDFox is an optimised RDF triple store and parallel Datalog reasoning engine developed at the University of Oxford. OST’s development team has applied state of the art techniques and deep expertise in computer science to exploit the full power of modern multi-core architecture.

RDFox scales extremely well with every additional processor core and brings, for the first time, mid-sized and large applications of Semantic Technologies into the reach of commodity hardware.

Datalog is a well understood rule language which allows for neat and elegant formulation of queries and views. With Datalog, one can encode specifications, regulations and policies given in natural language whilst maintaining a high correspondence to the original text. Datalog's clarity shortens development cycles and enormously increases the maintainability of implemented projects.

Developing solutions in Datalog requires less IT-expert skill and allows domain experts to formulate queries by themselves. Development cycles for data analysis tasks are drastically reduced. Datalog explanations for query answers show exactly how a certain answer was derived from the given data. This allows the domain expert to validate developed solutions and gives certainty about the obtained results.

Features

Capabilities

SPARQL query answering

Datalog and SWRL rules

OWL 2 RL ontologies

Extensions to Datalog
e.g. AGGREGATE, BIND, FILTER

Existential rules
e.g. SKOLEM, Negation as failure

Incremental reasoning

Explainability

Performance

Scalable and robust performance

Parallel reasoning

Highly optimised data structure via lock-free parallel inserts

Main memory, multi core architecture

Deployment

Supported on Windows, MacOS, Linux

REST API

Ready to deploy as a Docker image

Tested for Kubernetes

SSL security

Interested to find out more?

Glossary

ReasoninG

Reasoning is the work that RDFox does to apply the ontology to the data.

A part of RDFox, the reasoning engine, will look at the ontology, and apply it to the data to derive new implicit facts.

So, if our data says: John has a child Peter, Louise has a child Peter, Peter is male;

And if our ontology says: a male child is a son, if A has a child B, then B is a parent of A

RDFox’s reasoning engine will automatically derive that:

 John has a son Peter, Louise has a son Peter, Peter has a parent Louise, Peter has a parent John.

So if two users ask about Peter’s family in different legitimate ways, RDFox will be able to answer both, just like a human would.

Imagine if we could combine the speed of a computer with the intelligence of a human…

RDFox’s reasoning supports Incremental Reasoning.

This means that as new rules, or new facts, are added to (or removed from) the database, the consequences are automatically updated throughout the database until all the data is correct.

So, if a dependency chain of rules exists, the data will be updated automatically along the whole chain if need be.

RDFox is able to do all this thanks to years of research by our founders, who have proved, theoretically and empirically, that RDFox will generate the right results given the rules – at speed.

Explainability

The explain command allows the user to ask how a fact was derived.

It may have been part of the data before any rule was applied to it, or have been added as a conclusion of a sequence of rules. In either case, RDFox will explain.

This means that if a fact is derived that should not have been, the user can have a list of possible ‘wrong’ rules – rules which may seem correct but are not (e.g. we may have a rule to identify one’s colleagues as people working in one’s company. However, this will include oneself as one’s colleague – an easy mistake to make.)