Actively Learning ELI Queries under DL-Lite Ontologies

We show that ELI queries (ELIQs) are learnable in polynomial time in the presence of a DL-Lite ontology O, in Angluin’s framework of active learning. When initially provided with a conjunctive query (CQ) that implies the target ELIQ under O (in the sense of query containment), it suffices for the learner to only pose membership queries to the oracle, but no equivalence queries. The initial CQ can be obtained by a single equivalence query and is available ‘for free’ in case that O does not pose any disjointness constraints on concepts. Our main technical result is that every ELI concept has only polynomially many most specific subsumers w.r.t. a DL-Lite ontology, generalizing a recent result about homomorphism frontiers by ten Cate and Dalmau.

[1]  Jean Christoph Jung,et al.  Least General Generalizations in Description Logic: Verification and Existence , 2020, AAAI.

[2]  Guohui Xiao,et al.  Tractable Queries for Lightweight Description Logics , 2013, IJCAI.

[3]  Anni-Yasmin Turhan,et al.  Most Specific Generalizations w.r.t. General EL-TBoxes , 2013, IJCAI.

[4]  Ralf Küsters,et al.  Computing Least Common Subsumers in Description Logics with Existential Restrictions , 1999, IJCAI.

[5]  William W. Cohen,et al.  The learnability of description logics with equality constraints , 1994, Machine Learning.

[6]  Franz Baader,et al.  Computing the Least Common Subsumer w.r.t. a Background Terminology , 2004, Description Logics.

[7]  Michael Frazier,et al.  Classic learning , 1994, COLT '94.

[8]  Jean Christoph Jung,et al.  Reverse Engineering Queries in Ontology-Enriched Systems: The Case of Expressive Horn Description Logic Ontologies , 2018, IJCAI.

[9]  Boris Konev,et al.  Exact Learning of Lightweight Description Logic Ontologies , 2014, KR.

[10]  William W. Cohen,et al.  Learning the Classic Description Logic: Theoretical and Experimental Results , 1994, KR.

[11]  Jean Christoph Jung,et al.  Learning Description Logic Concepts: When can Positive and Negative Examples be Separated? (Abstract) , 2019, Description Logics.

[12]  Jens Lehmann,et al.  Concept learning in description logics using refinement operators , 2009, Machine Learning.

[13]  Diego Calvanese,et al.  The DL-Lite Family and Relations , 2009, J. Artif. Intell. Res..

[14]  Jean Christoph Jung,et al.  Logical Separability of Incomplete Data under Ontologies , 2020, KR.

[15]  Balder ten Cate,et al.  Conjunctive Queries: Unique Characterizations and Exact Learnability , 2021, ICDT.

[16]  Diego Calvanese,et al.  Tractable Reasoning and Efficient Query Answering in Description Logics: The DL-Lite Family , 2007, Journal of Automated Reasoning.

[17]  Franz Baader.  Least Common Subsumers and Most Specific Concepts in a Description Logic with Existential Restrictions and Terminological Cycles , 2003, IJCAI.

[18]  Daniele Nardi,et al.  An Introduction to Description Logics , 2003, Description Logic Handbook.

[19]  Andrea Calì,et al.  A general datalog-based framework for tractable query answering over ontologies , 2009, SEBD.

[20]  Pascal Hitzler,et al.  Efficient Concept Induction for Description Logics , 2018, AAAI.

[21]  Andrea Mazzullo,et al.  Learning Query Inseparable ELH Ontologies , 2019, ArXiv.

[22]  Dana Angluin,et al.  Learning Regular Sets from Queries and Counterexamples , 1987, Inf. Comput..

[23]  Magdalena Ortiz,et al.  Ontology-Mediated Queries from Examples: a Glimpse at the DL-Lite Case , 2019, GCAI.

[24]  Dana Angluin,et al.  Queries and concept learning , 1988, Machine Learning.

[25]  Boris Konev,et al.  A Model for Learning Description Logic Ontologies Based on Exact Learning , 2016, AAAI.

[26]  Ana Ozaki,et al.  Learning Description Logic Ontologies: Five Approaches. Where Do They Stand? , 2020, KI - Künstliche Intelligenz.