Research On Key Technologies Of Semantic Discovery Services For Web Manufacturing Resources

Networked manufacturing needs the wider share and uitilization of manufacturing resources. Manufacturing resource discovery on the web can provide the effective support for resource sharing. Aiming at the characteristics of web manufacturing resources and enterprise demands, the dissertation proposes the system architecture of semantic discovery service for web manufacturing resources. The domain-specific web search, information processing and application service are studied comprehensively. These technologies are employed to realize the web-extended resource services such as search & navagition of specialized resources, web resource repository, knowledge discovery and so on. The main work of this dissertation is dissertated in the followings:In the 1^st chapter, the research background and practical significance of web manufacturing resource discovery are introduced briefly. And the research content, development history and current advances related to web search and Semantic Web are also reviewed. An approach to manufacturing network-oriented resource semantic discovery service is put forward, and then the research content and architecture of the dissertation are presented.In the 2^nd chapter, the design objective of web manufacturing resource discovery service system (WMRD2S) is addressed firstly. After analyzing and comparing current systems, the logical hierarchy model based on the Semantic Web is brought forward. This model includes a domain ontology layer, which make semantics of web manufacturing resources clear to provide deep supports for resource acquisition, description and application. On the basis of this model, both the function model and architecture of WMRD2S are proposed. WMRD2S chooses J2EE as its development platform. In the 3^rd chapter, the correlation research of feature modeling theory of website topic and focused crawling technology is discussed in detail. To manufacturing network node recognition, a hybrid vector space model (HVSM) considering the semantic information of different texts, which is based on vector space model, is presented. Its key technologies cover the vector composition, weighting methods of vector elements, feature term selection and network search strategy. To facilitate application to identify manufacturing nodes, a Centroid-SVM classification algorithm is designed on the basis of HVSM. Finally, through contrast experiments of manufacturing node recognition under the semiclosed network enviorment and opened focused crawling, we carry on the experimental verification and performance evaluation of the above-mentioned theories.In the 4^th chapter, the technologies of semantic modeling as well as domain ontology construction of web manufacturing resources are discussed. In order to solve the complexity of ontology engineering and standardization requirements, a resource semantic hierarchy model (RSHM) is presented for domain ontology development. This model is divided into three layers: facet layer, concept layer and metadata layer. Furthermore, the realization of each layer is introduced. The extended operation of ontology and metadata is adopted in terms of practical applications. As to different metadata forms of manufacturing resources, a metadata extraction mechanism based on the matching pattern library is discussed.In the 5^th chapter, we analyze the characteristics of user-oriented resource discovery service, and propose the strategy of user-oriented resource service and semantic matching framework. The semantic hierarchy model similar to RSHM is employed to build user ontology of dynamic user demand information. The concept maintenance and weighting adjustment of user ontology take the system self-learning way. The semantic similarity measure of ontology concepts is taken as the measuring standard. A retrieval algorithm based on semantic similarity between resource ontology and user ontology is given. Finally, a personality push service of web resource is realized based on the user semantic ontology model.In the 6^th chapter, we take college resource sharing and rapid product development as the application background, and introduce the three typical examples about integration application services of WMRD2S. Through active web search and intelligent gathering of metadata for college instrument and equipment resources, effective resource navigation and information services are realized. The specific web search of part supplier information extends the resource acquisition and service functions of web parts library. The customize search makes for the personality search and push service for design knowledge and enterprise intelligence.The last chapter summarizes the achievements and innovations of this dissertation, and prospects the future work.