Epistemic Disjunctive Datalog for Querying Knowledge Bases


  • Gianluca Cima Sapienza University of Rome
  • Marco Console Sapienza University of Rome
  • Maurizio Lenzerini Sapienza University of Rome
  • Antonella Poggi Sapienza University of Rome




KRR: Description Logics, KRR: Computational Complexity of Reasoning, KRR: Logic Programming, KRR: Ontologies and Semantic Web


The Datalog query language can express several powerful recursive properties, often crucial in real-world scenarios. While answering such queries is feasible over relational databases, the picture changes dramatically when data is enriched with intensional knowledge. It is indeed well-known that answering Datalog queries is undecidable already over lightweight knowledge bases (KBs) of the DL-Lite family. To overcome this issue, we propose a new query language based on Disjunctive Datalog rules combined with a modal epistemic operator. Rules in this language interact with the queried KB exclusively via the epistemic operator, thus extracting only the information true in every model of the KB. This form of interaction is crucial for not falling into undecidability. The contribution provided by this paper is threefold. First, we illustrate the syntax and the semantics of the novel query language. Second, we study the expressive power of different fragments of our new language and compare it with Disjunctive Datalog and its variants. Third, we outline the precise data complexity of answering queries in our new language over KBs expressed in various well-known formalisms.




How to Cite

Cima, G., Console, M., Lenzerini, M., & Poggi, A. (2023). Epistemic Disjunctive Datalog for Querying Knowledge Bases. Proceedings of the AAAI Conference on Artificial Intelligence, 37(5), 6280-6288. https://doi.org/10.1609/aaai.v37i5.25773



AAAI Technical Track on Knowledge Representation and Reasoning