Approximate Reasoning in Description Logic

Content

Are you interested in pushing the boundaries of knowledge representation and reasoning? Join our project ApproxDL, where you will be developing new solutions for dealing with inconsistent Knowledge Bases (KBs).

In this project, we will focus on addressing a critical limitation in existing reasoners: their inability to handle inconsistent KBs effectively. Many real-world scenarios involve information that may conflict or contradict each other, and our goal is to enable reasoning even in the presence of such inconsistencies. In this project group (PG), you will implement a reasoner that can handle an inconsistent knowledge base.

Your group will employ three main strategies to handle inconsistencies: Intersection of All Repairs (IAR) aims to find a common subset of consistent information from all possible repairs, Conjunctive Query Reasoning (CQR) provides an answer for a query if it holds true in every repair, and Brave Reasoning (BR) offers an answer for a query if it holds true in at least one repair.

Roadmap

1. Read into DL semantics (EL, ALC) and OWL.
2. Implement subsumption type and property hierarchies (e.g. cat is subclass of mammal).
3. Implement the ApproxDL service.
4. Build an API for reasoning services.
5. Evaluate performance and result quality.

Requirements

• Very good Python skills.
• Experience in software development and with git flow.
• It is beneficial if you attended the Foundations of Knowledge Graphs lecture or a Logics lecture before.
• No fear of formal specifications, semantics, and formulae.

Course in PAUL

The PG will be available in PAUL (TBA).

Contact