Project Details
Projekt
Conservative Extensions in Ontology Languages: Beyond Description Logics
Subject Area
Theoretical Computer Science
Term
from 2014 to 2020
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 264279960
Ontologies are computer-understandable "dictionaries", which are usedin areas such as medicine, health-care, linguistics, businessprocesses and the Semantic Web. In the domain of Description Logic, awell-known and standardized family of ontology languages, a varietyof reasoning services have been developed, which make the design and useof ontologies in practical applications easier or at all possible. Inthis connection, the fundamental notion of conservative extensions hasturned out to be particularly useful because it underlies variousreasoning services that support refinement, reuse and versioning ofontologies, and allows to define notions of modularity suitable forontology engineering.Fortunately, for various description logics (and closely relatedmodal logics), the question whether a given ontology is a conservativeextension of another has turned out to be decidable. A variety ofmethods and characterizations have been developed to solve this kindof decision problem and ensure conservativity. In the proposedproject, we want to investigate to which extent the good computationalproperties of conservative extensions generalize to more expressivefragments of first-order logic. In particular, we will study theGuarded Fragment and its most important extensions, fragments with abounded number of variables, the Datalog+- family of ontologylanguages (also known as Tuple Generating Dependencies in databasetheory), and the recently suggested Unary-Negation andGuarded-Negation fragments.The aim of this project is twofold: on the one hand, we aim atunderstanding the reasons for the robust computational behavior ofdescription and modal logics with respect to conservativeextensions. Where is the frontier of decidability? Currently it isonly clear that it does not coincide with the decidability frontier forsatisfiability. On the other hand, in the past years a strong trendtowards using ontologies in data-intensive applications has becomeapparent, and description logics are suitable for those to only alimited extend because they are restricted to unary and binary relationsymbols. In order to solve this problem, the mentioned fragments offirst-order logic, in particular the Guarded Fragment and Datalog+-,are increasingly being used as ontology languages. The proposedproject is to develop an extensive theoretical understanding ofconservative extensions for these ontology languages, and thus delivera solid foundation for realizing the above mentioned reasoningservices and for the definition of notions of modularity.
DFG Programme
Research Grants
