Knowledge Base Systems

Built on research in Knowledge Representation and Reasoning, one of the oldest branches within AI.

A Knowledge Base System allows one to have a single representation of the information about a problem domain and to use it to solve a wide range of tasks.

Knowledge representation and reasoning is devoted to representing information about the world in a form that can be utilized to solve complex tasks. The IDP3 knowledge base system developed in our lab is based on (i) an expressive extension of first order predicate logic to capture in a natural way the knowledge about a problem domain (ii) a solver that can apply various inference methods in order to use the same knowledge to perform a wide range of tasks.

Detailed topics: 

Languages

Expressive logic constructs are needed to formulate many forms of knowledge. The current system supports first order logic extended with inductive definitions, types, arithmetic, aggregates and partial functions. For some problem domains, more is needed: constructed types, higher order logic, ...

Solvers and Inference Methods

Solving a wide range of tasks using a single logic based representation requires a powerful solver that supports many forms of inference, that can invoke existing solvers (constraint programming, mixed integer programming, ...) and that can interact with a procedural environment.

Applications

Many search problems have a succinct representation within IDP3. The system has been successfully applied on a number of machine learning and data mining problems.

People: 
  • Marc Denecker
  • Maurice Bruynooghe
  • Gerda Janssens
Events: 
Software: 
  • IDP: A model expansion system for an extension of classical logic
Selected publications: 
  • Broes De Cat, Bart Bogaerts, Maurice Bruynooghe, Marc Denecker. "Predicate Logic as a Modelling Language: The IDP System." CoRR abs/1401.6312. 2014. PDF
  • Marc Denecker, Eugenia Ternovska. "A logic of nonmonotone inductive definitions" in ACM Transactions on Computational Logic, volume 9, issue 2. 2008. PDF
  • Maurice Bruynooghe, Hendrik Blockeel, Bart Bogaerts, Broes De Cat, Stef De Pooter, Joachim Jansen, Anthony Labarre, Jan Ramon, Marc Denecker, Sicco Verwer. "Predicate logic as a modeling language: Modeling and solving some machine learning and data mining problems with IDP3." in Theory and Practice of Logic Programming. 2014. PDF
  • Maarten Mariën, Johan Wittocx, Marc Denecker, Maurice Bruynooghe. "SAT(ID): Satisfiability of propositional logic extended with inductive definitions." in Lecture notes in computer science, volume 4996, pp. 211-224. 2008. PDF
  • Johan Wittocx, Marc Denecker, Maurice Bruynooghe. "Constraint Propagation for First-Order Logic and Inductive Definitions." in ACM Transactions on Computational Logic, volume 14, issue 3, 2013. PDF
  • Joachim Jansen, Gerda Janssens, Albert Jorissen. "Compiling input∗ FO(·) inductive definitions into tabled Prolog rules for IDP3." in Theory and Practice of Logic Programming, volume 13, issue Special Issue 4-5, pages 691-704, 2013. PDF
  • Broes De Cat, Marc Denecker, Peter Stuckey, Maurice Bruynooghe. "Lazy model expansion: Interleaving grounding with search." in The Journal of Artificial Intelligence Research, volume 52, pages 235-286, 2015. PDF
  • Hanne Vlaeminck, Joost Vennekens, Marc Denecker, Maurice Bruynooghe. "An approximative inference method for solving ∃∀SO satisfiability problems." in The Journal of Artificial Intelligence Research, volume 45, pages 79-124, 2012. PDF
  • Johan Wittocx, Maarten Mariën, Marc Denecker. "Grounding FO and FO(ID) with bounds." in The Journal of Artificial Research, volume 38, pages 223-269, 2010. PDF