A Similarity Measure for the Description Logic EL with Unfoldable Terminologies

Abstract

Description Logics (DLs) are a family of logicbased knowledge representation formalisms, which can be used to develop ontologies in a formally well-founded way. The standard reasoning service of subsumption has proved indispensable in ontology design and maintenance. This checks, relative to the logical definitions in the ontology, whether one concept is more general/specific than another. When no subsumption relationship is identified, however, no information about the two concepts can be given. This work presents a new notion of semantic similarity which stems from the known homomorphism-based structural subsumption algorithm. The proposed similarity measure computes a numerical degree of similarity between two ℰℒ concept descriptions despite not being in the subsumption relation.

Similarity-based relaxed instance queries

“…The measures described in [15,16] fulfill most of these properties. The parameterizable similarity measures from [15] additionally allow users to calibrate the measure to fit their expectations.…”

“…Moreover, many structural CSMs used in practice, like those presented in [15,16], are based on a recursive computation of the similarities between concepts at the same role-depth. This means that the computation of the similarity only recursively visits existential restrictions, until one subconcept has no further existential restrictions.…”

“…Furthermore, CSMs are used in ontology alignment algorithms [12]. For DLs there exists a whole range of CSMs , which could be employed for the task of answering relaxed instance queries [13][14][15][16]. In particular the CSMs generated by the framework described in [15] allow users to specify which part of the vocabulary used in their knowledge base is to be regarded more important when it comes to the assessment of similarity of concepts.…”

Similarity measures for FL₀ concept descriptions from an automata-theoretic point of view

“…One may argue to adopt an alternative methods for calculating a similarity degree based on both directions of the subsumption degree, such as, adopting the multiplication and adopting the root mean square (rms) [2]. Unfortunately, these methods give unsatisfactory values for the extreme cases such as a concept A and the top concept .…”

“…When no subsumption relationship between two concepts is identified, no classification relationship between the two concepts can be given. This lack leads to an introduction of the non-standard reasoning service of concept similarity measure as in [2], [3] and [4].…”