Integrity Constraint Validation In OWL Ontologies

Tim Berners-Lee has put forward the concept of Semantic Web in1998, in which datahave been given the computer-readable semantics for the human-computer interaction.Ontology is the core of the Semantic Web and is used for representing the conceptual modelof Semantic Web data. Web Ontology Language (OWL) is the endorsed standard ontologydefinition language for the World Wide Web Consortium (W3C). As the logical foundation ofOWL, Description Logics(DLs) have expressive languages and provide sound and completereasoning algorithms for standard reasoning tasks, such as concept satisfaction, consistency ofontology, and ontology debugging etc..In data-centric applications, besides avoiding the logical conflicts in ontology data, usersprefer to the integrity constraint validation in ontology, which means the ontology data shouldcomplete and satisfy certain conditions for user-requirements. Integrity constraints wereoriginally proposed in database and artificial intelligence knowledge representation languagesto guarantee the legal states that are considered acceptable by knowledge bases. Integrityconstraints can be added into the ontology to guarantee the integrity of ontology. For example,in the development of information systems, integrity constraints are required to represent theapplication requirements in the ontology. Furthermore, in the translation from a databasemodel to an OWL model, they are also required to identify the ontology elements andrestrictions on these elements. By doing the validation of integrity constraint satisfaction, it ispossible to guarantee the integrity of ontology data. Thus, how to validate the integrityconstraints in ontology is an essential issue in the ontology engineering.There are four issues should be considered in integrity constraint validation. Firstly, themain challenge of modeling integrity constraints in ontology is the difference in semantics.The integrity constraint validation should adhere to the closed world assumption (CWA),whereas the standard semantics of OWL follows the open world assumption (OWA). Thestandard OWL semantics makes data in the semantic Web faithfully capture the information inthe real world. However, due to the appearance of uncertain information, it is difficult tomodel such constraints under the OWA. Therefore, we are required to model the integrityconstraints under the CWA. Secondly, according to the application requirements, it isnecessary to gurantee that the ontology data should be minimized w.r.t. integrity constraints.Thirdly, we should consider the combination of standard reasoning and integrity constraintvalidation. Since the ontology has the ability of inferring implicit information, the standardreasoning is also essential and required to be considered during the integrity constraintvalidation process. Last but not least, we also need to consider the efficiency of integrity constraint validation. At the basis of the ensurance of integrity constraint validation, how toimprove the efficiency of integrity constraint validation in an ontology is an important issue.Especially, when face to the scalable Web data, the efficiency may directly affect the variousapplications in ontology engineering.With the analysis of existing works, we study integrity constraint validation from thefollowing three aspects, contains the semantics of integrity constraint in ontology, thecombination of standard reasoning and integrity constraint validation, and the optimization ofintegrity constraint validation.1. Firstly, we model the semantics of integrity constraints in OWL ontology using groundedcircumscription and weak unique name assumption in this paper. By restricting theextension of predicates in integrity constraint axioms, we define the semantics ofintegrity constraint axioms which is under closed world assumption. Furthermore,according to the minimal idea of grounded circumscription, the satisfaction of integrityconstraint is also defined. Finally, in order to get more intuitive sense of integrityconstraint semantics, we additionally apply the weak unique name assumption. Underthis semantics, the reasoning algorithm may well combine the standard reasoning underthe open world assumption and integrity constraint validation under the closed worldassumption, more appropriate to the user requirements.2. In order to well combine the standard reasoning in description logic knowledge bases andintegrity constraint validation, we validate the integrity constraint by modifying tableaualgorithm and introducing negation as failure respectively in this paper. The formermethod mainly focuses on modifying the expansion rules in tableau algorithmsaccording to the defined integrity constraint semantics. The latter method firstlypreprocesses integrity constraint axioms with negation operators and then validate theintegrity constraint knowledge bases entailments which using the tableau algorithm.Each of these methods has its advantage: the former has high efficiency and the lattermethod is a universal approach which is suitable for the most of description logiclanguages.3. We optimize the integrity constraint validation by reformulating and modularization.Firstly, we consider the validation of integrity constraints for tractable DL-Lite ontology.Using reformulation technology, the satisfaction of integrity constraint axioms can bereduced to the query answering over databases. By doing this, we can directly using thequery optimization mechanisms to optimize the efficiency of integrity constraintvalidation, which may more appropriate to the scalable ontology data. Additionally, weuse the modularization to extract the modules, which are logically related with integrityconstraint axioms, from scalable ontology data, and further check the satisfaction ofintegrity constraint axioms in these modules. In this way, it is no longer necessary to dothe integrity constraint validation in the whole ontology, and get the same results for the integrity constraint validation.