Title: Distributed Reasoning in the Web of Data

Duration: 2 years

Research Area: AI Algorithms and Methods, Knowledge Representation and Engineering

Since its invention in the early 2000s, the idea of creating a Semantic Web has gained a lot of traction. Standards for expressing and sharing linked data like the Resource Description Framework (RDF) have been created and the Linked Open Data Cloud containing publicly available information following this standard keeps growing. Logical relations between different resources can be expressed in RDF Schema (RDFS) or in the Web Ontology Language (OWL), industry and public sectors (see also linked open vocabulary) provided reusable vocabularies following these specifications.

While all these can be seen as indicators for a success of the Semantic Web, there are also aspects which are not fully realized.  Following the OWL or RDFS vocabularies, reasoners can draw conclusions from the data at hand, but these conclusions and the way they were obtained can currently not be shared through the Web as there is no standard for doing so.  If reasoners follow other logics, for example rule-based approaches, they mostly do not even share their derivation rules in an Semantic Web standard (despite the fact that this standard exists).

As a consequence, Semantic Web data is mostly processed locally and the lack of standards which are actually used hinders the Semantic Web from using its full potential: In a Web of Data, reasoners should be able to rely on each other’s results, it should be possible to check reasoning results in case of doubt, and meta information like the provenance of a fact should be accessible in a machine-readable way. An infrastructure like that would allow distributed reasoning and proof-checking.

Aims

The aim of the research project is to investigate how different logics and proofs can be expressed, shared and checked through the Semantic Web. Important aspects here are how to reference sources of data and derivations and how to extend the existing standards to allow for meta-statements. Of course these possible extensions need to be semantically sound.

Problem

Our project aims to answer the question on how the different technologies and standards in the current Semantic Web can be used in combination. One of the problems is that it is not always clear how standards or frameworks relate to each other. Another problem is that it is not possible to make statements about statements and extending logics to allow these kinds of expressions is far from trivial.

Practical example

In the term of this project we have been and are still active in the creation of different standards for the semantic Web like for example RDF-star – a framework allowing users to use statements as terms (provenance), Notation3 Logic (https://www.w3.org/community/n3-dev/ ) – a framework supporting rules, graph terms but also proofs, and RDF-Surfaces – a framework to express First Order Logic in the Web.

Technology

In this project we mainly use implementations of different translations between frameworks/formalisms. We for example implemented a translator from Notation3 Logic to existential rules and from SPARQL to Notation3 Logic. We furthermore plan to implement proof checkers for proofs written in Semantic Web frameworks.

Outlook

The research project will connect the different branches of the Semantic Web which currently work separately. If it is possible that different parties share their derivation results but also their knowledge about the trustworthiness of different sources, reasoning tasks can be optimised.

Publications

• https://www.w3.org/2021/12/rdf-star.html
• https://link.springer.com/chapter/10.1007/978-3-031-45072-3_5
• https://ceur-ws.org/Vol-3632/ISWC2023_paper_469.pdf

Team

Lead

• Prof. Dr. Sebastian Rudolph

Team Members

• Dr. Dörthe Arndt