Research On Semantic Information Management And Retrieval Inside Resource Sharing Networks

Along with progress in the information technology, there emerge some new research hotspots about information processing on Internet, such as Semantic Web (SW), Peer-to-Peer (P2P), Web Services (WS) and so on. These scientific models attract more attentions of global research communities and bring forth many solutions for some important problems. Such techniques share many common ideas and principles, and they can learn from each other as well as keep their characteristics, therefore, many researchers begin combining different expertise of such models together to resolve hard problems. In this paper, the sharing infrastructure based on P2P is built. It exploits the SW techniques to represent and encode the resources, and provides the mechanism to access such semantic information, consequently to promote sharing for data resources and service resources.Firstly, the SemreX platform is designed and built to promote literatures sharing under wide area network. It takes the P2P and SW as the basement, and provides operation functions such as metadata extraction, semantic integration and semantic retrieval. A serial of ontologies or metadata models are given, e.g. Literature ontology, peer profile and literature taxonomy. How to utilize these ontologies to encode semantic information is introduced, too. To satisfy requirements of local repository management, the active storage and the duplicate detection policy are presented. Furthermore, to support the semantic retrieval, an integrated approach for indexing, storing and retrieving domain knowledge is proposed. It indexes the domain knowledge into inverted list of Lucene by paring ontology, and supports corresponding mechanism to retrieve knowledge. This approach can obtain the advantages on indexing and retrieving the domain knowledge.Secondly, a literature retrieval model is designed and exercises in the SemreX. The model not only considers how to retrieve an entity, but also evaluate its importance which is as the base of retrieval ranking within a domain. For this, the adaptive PageRank is introduced to perform literature evaluation in real domain; and the experiment shows that after adaptive modification, links analysis can also work well under the SW condition. Coming with document retrieval, a simple method and an upgraded vector spread activation model are proposed. They can combine semantic information and content of a document to assist the document retrieval, and improve the precision on ranking results. Besides, some methods are suggested to merge literature instances and support the similar literature recommendation.Thirdly, the SemreX sharing network is extended as Service@SemreX to support service resources management. The architecture takes the Web Service as a kind of knowledge and stores it locally to facilitate global sharing. In order to improve service location efficiency in the network, a semantic-overlay driven discovery means and the step-based discovery process are proposed. Then to overcome the poor efficiency with logic matching, the service selecting skill based on similarity computation at local peer is given. In addition, a context model which serves as the base of the service recommendation policy is designed; it can evaluate the importance of entities inside a service network and then recommend the services against their importance. According to the simulation, above-mentioned framework can satisfy the design target, as well as support effective service resources searching and better service recommendation.At last, to overcome the existent problems in service agreement management, the service agreement modelling based on the ontology technique and the corresponding template repository management approach are explained, respectively. By modelling the service agreement template in ontology, the concepts and parameters inside agreements can be defined in unified, rigorous and semantic-rich manner. Then how to semantically organize and build service agreement template repository to support component re-usage is introduced. During this stage, a step-based service agreement matching algorithm is given; it can make matching against service request description and QoS requirement. Besides, the template generating algorithm can born new candidate templates from the agreement knowledge repository to satisfy the custom's requests. Exploiting OWL to express service agreement can make a new agreement language, and it can easily be integrated with the agents to realize the management automation. Moreover, the policies and the formalized statements equipped with SWRL make a good foundation for reasoning. Finally, considering the QoS model of the semantic web services is not complete, a QoS metric model is devised to reinforce it.This paper has analyzed the sharing network concept and concluded the resources management and searching mechanism inside sharing networks. Then by combining application background, it has studied semantic information management in literatures sharing network, and services network, respectively. It has also developed a serial of technical approaches and measures to manage and share two importance resources as literatures and services on Internet.