Research On The Approaches To Combining Ontologies And Rules In The Semantic Web

The Semantic Web is an extension of the current World Wide Web, which helps computers"understand"the information on the Web and automatically process the information by various established standards and technologies, so that it better enables computers and human beings to work in a cooperative way. The keys for the realization of the Semantic Web are to add machine-readable semantic annotations to the existing information on the Web, to construct ontologies in order to share the precise definitions of web resources and then explore knowledge representation technologies for automated reasoning on these web resources.At present, the layers of the Semantic Web stack up to the ontology layer have reached certain degree of maturity, evidenced by the bunch of published W3C Recommendations. The next step aims at sophisticated representation and reasoning capabilities of the Rules, Logic and Proof layers of the Semantic Web. Due to the fact that spanning from ontology layer to logic layer is of considerable intractability, the integration ontologies and rules as an intermediate step has become a central hot topic in the Semantic Web context. However, three issues arise when combining a rule language with an ontology language: i) from the logical point of view: the standard Web Ontology Language is based on Description Logic, whereas, the existing proposals for a rule language for use in the Semantic Web originate from Logic Programming; ii) from the semantic point of view, Description Logics adhere to the Open-World semantics, while rules are typically interpreted under Close-World Semantics; iii) from the reasoning point of view, reasoning in a formal system obtained by integrating an ontology and a rule component maybe not be a decidable problem which is very hard to handle.Moreover, most of the existing approaches to combining ontologies and rules focus on: i) the crisp information, they hence are lack of capability in the case where uncertainty and vagueness information populate; ii) logical rules, they hence turn out to be problematic and not appropriate in reality, for instance, where maintenance and reuse are important issues. Therefore, both combining ontologies and rules with fuzzy facility, as well as combining ontologies and production rules are challenging and meaningful.Based on the analysis of the related research and existing methods, this thesis performed researches on the combination of terminological default theory and rules, combination of intuitionistic fuzzy description logics and rules, combination of ontologies and production rule systems, expert finder based on web data et al. The main results obtained by this thesis are summarized as follows:1) We have in-depth and in-width investigated state-of-the-art approaches to combining ontologies and rules from a theoretical point of view and a practical point of view, respectively. Firstly, in respect of theoretical approaches, we classify them into three approaches: a) loose coupling approaches allow a rather narrow interface between the semantics of the ontology component and rule component, and the interface defined based on logical entailment; b) tight coupling approaches provide a rather broad interface between the components, and both ontologies and rules are interpreted using common interpretations; c) embedding approaches provide a full integration under an uniform logic in order to realize a full integration. We give the definitions of syntax and semantics of approaches for each class, and introduce the representative systems and analysis their properties, then distinguish the difference between these approaches. Secondly, in respect of practical approaches, we describe three combinations of ontologies and rules: combining RDF with rules, combining OWL with rules and combining F-logic with rules. By analyzing the characteristics of these three ontology languages, we have respectively identified the features of combination approaches and introduced the tools at hand used to achieve these approaches. The research on the approaches to combining ontologies and rules provides us more in-depth understanding of the theoretical values brought by the approaches, and it makes us more aware of the existing problems. In this way, it's helpful for us to realize the aim of seamless combination of ontologies and rules, and live up with the realistic requirements from the Semantic Web area.2) The experience in building practical applications has revealed several shortcomings of most of research work in the context of the integration of description logics with Datalog rules or its non-monotonic extensions, because they frame a severe syntactical limitation on DL component that no default negative DL predicate occur in rules. Obviously, this syntactical restriction makes hybrid rules lose the facility to handle default assumptions and commonsense reasoning and modeling exceptions. To address the above limitation, we have proposed a combined system DLdlog as a family of hybrid rules especially featuring a capability for negation dl-atoms. We have integrated a decidable description logic (here, it ranges from ALC to SHIQ) and a finite set of defaults into logic programs and allowed negative dl-atoms like rule atoms to occur behind"not"in the body of hybrid rules, and treated in nonmonotonic way by NMD-Model rather than the NM-Model of DL+log. In this way, negative dl-atoms are treated in nonmonotonic way under NND-semantics. We have defined the syntax and semantics of the DLdlog knowledge bases; we have given an algorithm for satisfiability of restricted form of DLdlog knowledge bases, we have discussed the decidability and data complexity of DLdlog knowledge bases. We have proved that DLdlog enriches the expressivity of DL+log while NMD-Semantics in DLdlog formalism still remains faithful to NM-semantics in DL+log.3) In the realistic world, there extensively exists fuzzy knowledge other than precise knowledge. However, most of approaches of combining ontologies and rules are lack of the capability of expressing and reasoning fuzzy knowledge. To address this limitation, we have proposed IFDLflog, a system that combined by intuitionistic fuzzy description logics and intuitionistic fuzzy disjunctive Datalog programs. This formalism strengthens the ability of DL+log for the inherently imprecise applications. We have provided the abstract syntax and IFNM-Semantics of IFDLflog knowledge bases. We have defined logics operators acting on the truth values of L2 for negation, conjunction, disjunction and implication. These operators will be flexibly adapted in line with requirements from different scenarios. We have defined a satisfaction function to allow rules to be (not) satisfied to a certain degree. Moreover, an aggregator is chosen in some cases where rule have preferences. We have presented many semantic properties of such combination. We obtained a decidable deciding consistency and answering queries in the size of data by exposing some suitable assumptions. IFDLflog is qualified with fuzzy features while its IFNM-Semantics still remains faithful to still remains faithful to NM-semantics in DL+log.4) According to the requirements from application scenario, logical rules have feedbacks on expressing optional or action rules. To address this limitation, we have presented a weak coupling of Production Rule Systems with Description Logics. The basic idea behind this coupling is that the facts in the working memory are encoded into assertions in the ABox of the DLR-KB and each production rule is encoded into the form of implication. We have then given the algorithm for the query answering and shown how the algorithm works by an intuitive example. Furthermore, we have discussed the complexity of the algorithm. The advantage of our framework is the definition of dynamic axiom as an implication, which serves the basis of a formal model of PR execution over both ground memories (the ABox) and memory schemas (the TBox). Such a formal model can be used to verify the correctness of PR sets, whether or not they are confluent. Moreover, such a model is suitable to the algorithm proposed that identifies internal inconsistencies. The framework also turns out to be an optimal trade-of between an expressivity and a computational complexity of the reasoning tasks.5) To address the problems existing in expertise management, we have proposed an ExpertFinder prototype composed of four components, each of which has been described in more detail. The prototype demonstrates it possible to exploit the web of data for world-wide experts and expertise identification leveraging the semantic web technological stack, including RDF, reasoning, SPARQL, thesauri and rules over ontologies. The proposed prototype describes raw data from different open communities using standard vocabularies such as FOAF, SIOC, DOAP and SKOS. In this way, data from independent sources can be integrated into a single model. Data quality is a critical issue which is addressed by two smushing strategies. The aim is to figure out and reduce identical instances under different identities. We have evaluated the smushing strategies and we showed promising results that make us confident about the quality of the data used for the expertise inference process. An experiment was carried out to measure the accuracy of the results of the smushing process. Feedback was requested from the people who were identified as members of the at least four of the communities at hands. The amount of response was insufficient to extract significant conclusions, but the individual responses were generally positive and supportive. We have put effort in efficiently integrating inference techniques based on a set of custom rules, which are executed on top of all data to reveal expertise evidences of people for each topic in a given domain. We used some mathematical functions to aggregate expertise from different evidences in order to create a consistent and reliable profile. At the same time, the success of this system within an enterprise environment has demonstrated the effectiveness of the related methods. The research results of this thesis including the approaches to combining ontologies and logical rules to deal with fuzzy knowledge, combining ontologies and production rules and Expertfinder based on the Web Data will enrich and push forward the studies of the related areas in both theoretical and technological aspects.