DL-Lite Ontology Revision Based on An Alternative Semantic Characterization

Abstract

Ontology engineering and maintenance require (semi-)automated ontology change operations. Intensive research has been conducted on TBox and ABox changes in description logics (DLs), and various change operators have been proposed in the literature. Existing operators largely fall into two categories: syntax-based and model-based. While each approach has its advantages and disadvantages, an important topic that has been rarely explored is how to achieve a balance between syntax-based and model-based approaches. Also, most existing operators are specially designed for either TBox change or ABox change, and cannot handle the general ontology revision task-Given a DL knowledge base (KB, a pair of a TBox and an ABox), how to revise it by a set of TBox and ABox axioms (i.e., a new DL KB). In this paper, we introduce an alternative structure for DL-Lite, called a featured interpretation, and show that featured models provide a finite and tight characterization to the classical semantics of DL-Lite. A key issue for defining a change operator is the so-called expressibility, that is, whether a set of models (or featured models here) is axiomatizable in DLs. It is indeed much easier to obtain expressibility results for featured models than for classical DL models. As a result, the new semantics determined by featured models provides a method for defining and studying various changes of DL-Lite KBs that involve both TBoxes and ABoxes. To demonstrate the usefulness of the new semantic characterization in ontology change, we define two revision operators for DL-Lite KBs using featured models and study their properties. In particular, we show that our two operators both satisfy AGM postulates. We show that the complexity of our revisions is Π P 2-complete, that is, on the same level as major revision operators in propositional logic, which further justifies the feasibility of our revision approach for DL-Lite. Also, we develop algorithms for these DL-Lite revisions.

DL-Lite Contraction and Revision

“…We will focus on the scenarios (1) and (2). In managing DL KBs, scenario (3) is less common and it has been handled in (Wang et al, 2015) through a similar approach. 8…”

“…In dealing with changes to DL KBs, many are like us considering it as a belief change problem (Qi et al, 2006;Qi & Du, 2009;Qi et al, 2008;Ribeiro & Wassermann, 2009;Wang et al, 2015). Qi et al (2006) gave a weakening based approach for revising ALC KBs.…”

“…The idea is to weaken the TBox axioms until all inconsistencies are resolved. Qi and Du (2009) and Wang et al (2015) adapted Dalal's (1988) and Satoh's operators (1988) respectively for revising DL KBs. The main issue with these works is that their revision postulates are not formulated appropriately to capture the rationales of incoherence resolving.…”

“…To deal with changes to DL KBs, it makes sense to take advantage of the existing techniques in belief change. In fact, many have investigated contraction and revision over DL KBs (DL contraction and DL revision for short) (Qi et al, 2006;Qi & Du, 2009;Qi et al, 2008;Ribeiro & Wassermann, 2009;Wang et al, 2015).…”

“…None of them provides representation theorem for their contraction or revision function except the work by Ribeiro and Wassermann (2009) which inherits the representation results from a more general work (Hansson & Wassermann, 2002). 1 In defining DL revision, many did not appreciate its incoherence resolving nature, as their revisions cannot guarantee coherence of the outputs (Qi et al, 2006;Ribeiro & Wassermann, 2009;Wang et al, 2015). Qi and Du (2009) appreciate the incoherence resolving nature, but the postulates they provided for capturing properties of their revision function are not formulated appropriately to capture the rationales behind incoherence resolving.…”