The Global Standard of Semantic Reasoning

Semantic reasoning, also known as rules-based AI, is the process of enriching databases with valuable insights by leveraging expert knowledge and contextual information to infer new data. The flexibility of a knowledge graph then enables RDFox users to ask impactful questions and extract answers to suit their business’s needs.

RDFox stands as the world's most powerful enterprise reasoning engine, delivering the exceptional speed and utmost accuracy required in industry. This unrivalled semantic reasoning is the result of decades of pioneering research at the University of Oxford that have now culminated in our proprietary knowledge graph database and reasoner, RDFox.

Prof. Ian Horrocks, considered by many to be one of the fathers of the semantic web, alongside the rest of the founding professors have set the global standard for reasoning. Having developed many pioneering techniques over the years, Ian was awarded the Lovelace Medal for his lifetime contributions to the field in 2020.

What is Semantic Reasoning?

Semantic reasoning, also referred to as semantic inference or rules-based AI, is a powerful technique that enhances data processing and decision-making in businesses. By applying a predefined set of rules, semantic reasoning automatically incorporates new data into a database, expanding the knowledge base and enriching insights.

These rules, defined by the user, leverage the existing data to generate logical consequences, identify patterns, and derive more meaningful relationships within the dataset, revealing deeper insights that go beyond the explicit data representation.

The intuitions behind knowledge graphs and reasoning

Rules-Based AI vs. Machine Learning

Unlike machine learning, rules-based AI is based in logic as opposed to statistical models, ensuring 100% accuracy and explainability in all results. This unique advantage empowers RDFox users to address industry’s most critical challenges, relying on the auditable truth provided by RDFox as a trusted source of information when it matters most.

Despite being frequently positioned as contrasting forces, there is much to be gained by combining rules-based AI and machine learning, as the former provides speed and accuracy to offset the drawbacks of the latter. Some of today’s most innovative solutions blend the two technologies, and given their success, it’s a certainty that tomorrow’s will too.

Ask the Profs. Semantic Reasoning vs. ML

Why use Semantic Reasoning?

The purpose of semantic reasoning is to supercharge a database, making knowledge retrieval faster and more efficient while providing enhanced and contextualised insights into the data. By inferring new patterns and relationships, queries can be drastically simplified, reducing their execution time by orders of magnitude. These new relationships can also be created such that more valuable questions can be asked of the data, allowing RDFox users to extract more meaningful answers.

Such an improvement in speed and functionality has proven transformative to many RDFox clients, now able to create solutions that were never before possible. It doesn’t stop there though. Reasoning enables all results to be explained down to the rules and root facts that lead to them; it assists data integration, transformation, and management; and acts incrementally—staying up to date even as new data is added, without the need to a restart.

All of this in combination allows businesses to get the most out of their data, extracting more value, minimising downtime, and empowering informed decision making.

How rules impact queries

What knowledge representation languages does RDFox support?

RDFox supports OWL (Web Ontology Language—the community standard), SWRL, and an extension of Datalog—now an extremely powerful and expressive rules language. The majority of RDFox users favour Datalog for its intuitive syntax and extensive functionality including negation, aggregation, built-in functions, and so much more. See the documentation for the full list.

See the documentation