Research On Knowledge Service For The Semantic Web

WWW is the largest repository of information that human beings have ever made. The URI andHTTP construct a globally accessible information space, in which every object has its own uniquename. Compared with the database technology which constraints data sets with global consistency,the Web technology facilitates the information space becoming universal by only maintaining localconsistency. With the increasing demand of resource sharing, WWW is gradually becoming a sup-porting platform of information sharing. And the Semantic Web, based on Logics and specifications,will have a leading role in future Web information systems.The Semantic Web is to add a machine-readable meaning to Web pages, use ontologies for a pre-cise definition of shared terms in Web resources, make use of Knowledge Representation technologyfor automated reasoning of Web resources, and apply cooperative agent technology for informationprocessing on the Web. Knowledge Service in the Semantic Web studies how to share, organize,manage and merge Web resources in an open and scalable semantic-interconnection environment,to provide a dynamic, cross-platform and virtual community oriented interaction patterns, and on-demand knowledge services to support cooperative working, problem solving and decision support.This thesis mainly focuses on two key problems in Knowledge Service, i.e., the service dynamicsmodelling and formal resource organization. In this dissertation, I will describe four contributionsmade to the field of Knowledge Service in the Semantic Web:1. Improve the DDL formalism. In DDL, a formal meaning is assigned to the interaction be-tween services and the environment. Thus DDL bridges the gap between the static reasoningon content and the dynamic reasoning on services in an inherent, consistent and seamless way.As a formalism, DDL combines description logics and action formalisms to support servicemodelling and reasoning. We define a set of action constructors, on which complex actions canbe constructed. We also discuss the consistency of action, propose a Tableaux-based algorithmfor action consistency checking, and define action executability and projection. Based on De-scription Logics, DDL enables a better view of service description and performance, facilitatesinteroperability between services by providing them with shared knowledge, and provides alogical foundation for service applications, esp. on Semantic Web.2. Propose Logic Programming based action representation and reasoning support. Based on athorough analysis of action and the world model, the interaction between them embodies thekey feature of a transition system. Since Logic Program is an e?cient way for the transitionsystems, a reasoning support for DDL is given by translating actions into logic programs,and thus, the reasoning tasks in DDL can be transformed into computing answer set of logicprograms. The transformation can support the service process reasoning on ontologies andrules, resolve the frame problem by utilizing nonmonotonicity and procedural attachments oflogic programming, and give a formal basis for service reasoning on Semantic Web by reusingalready available algorithms and engines on logic programs.3. Propose an interval-based knowledge model and query language for temporal information.Usually, the performance of services will generate timestamped historical information aboutthe change of the environment. After a thorough analysis of the pros and cons of temporalER models and temporal description logics, we propose an interval-based knowledge modelfor time-varying data modelling in order to keep a better balance between formal semanticsand expressivity. The model is composed of a temporal knowledge schema, temporal integrityconstraints and an interpretation structure. By referring to temporal ER models, a group ofexpressive temporal constructs in the schema are defined to capture the temporal features ofconcepts, attributes and relationships;a group of rules are defined to enrich expressivity of theschema;by referring to temporal description logics, a formal meaning is given to the modelvia an interpretation with model-theoretical semantics. A high-level query language is definedwith temporal Boolean expression, temporal selection and temporal projection, which providean e?ective way for temporal data manipulation.4. Design a model for service-oriented and ontology-driven knowledge management. By makingthe semantic spaces and the knowledge space on Web resources, this model is proposed toexplore important aspects of service-oriented and ontology-driven knowledge management onthe grid. In the semantic layer, grid ontology services introduce domain ontologies, buildonto-data mappings, and form semantic spaces on local resources. The semantic spaces hidethe heterogeneities among local data, support semantic-based data query, and thus providea basis for resource discovery. In the knowledge spaces, grid ontology mediation servicesreorganize these semantic spaces, mediate the semantic heterogeneity among them, and forma virtual group in which the members know each other correctly based on ontology mappings.This model can construct a dynamic and scalable semantic-interconnection environment, helppeople acquire, publish, share and manage various Web resources.Based on the above research works, we discuss the issues of a multi-agent knowledge service envi-ronment. In the semantic layer, semantic access agents introduce domain ontologies and managelocal resources by the ontology-to-data mappings. In the knowledge layer, gateway agents buildontology mappings to bridge the semantic gaps between ontologies, connect semantic spaces andreorganize them into VHVC. In the service layer, we discuss the three abstract levels of services oractions, the desires and intentions of agents, and the combination of them in service compositionson a multi-agent environment.