Study On Distributed Knowledge Base System Based On P2P Network Infrastructure

In the passed couples of decades, communities focusing on the Semantic Web devoted much effort to penetrate this aspect and deal with issues within it. Information integration and query answering over autonomous, heterogeneous data sources are two typical missions relevant to the Semantic Web, which aims to provide a rich environment for data sharing and reusing across applications.This thesis introduces a novel approach, which aims to provide an integrated environment for users to accomplish information integration and query answering. It establishes a peer-to-peer network infrastructure, where each peer holds a substantial number of ontologies as its local resource and exchanges the information about these ontologies with others so that autonomous ontologies can be dynamically discovered by remote. Taking these ontology-related information into account, users on a certain peer can integrate arbitrary ontologies (either residing at local or at remote) to build an information integration system in a decentralized way and perform deductive queries over this system.Besides, this system employs the idea of distributed computation to optimize the reasoning tasks greatly, which in other words, queries over an integrated system can be performed concurrently and collaboratively on several peers.For issues related to information integration, it employs the progresses from several fashion research efforts to establish a model for heterogeneous data sources integration. A group of global ontologies builds up a decentralized and networked structure to define the mappings between these autonomous data sources by using rule expressions. Users on the top have a view of the integrated information system and perform reasoning tasks over it.For issues related to query answering, it takes especially the distributed query answering into account and proposes an mechanism to perform distributed computation for deductive reasoning. As queries are concurrently executed on several peers, it balances largely the computation load and optimizes the performs to the user.After a concrete illustration for the system both from theoretical and practical point of views, we concluded this thesis with an outlook towards further challenges relevant to the information integration and query answering.