Network Resources Annotation Based On Chinese FrameNet Ontology

Tim-Berners first proposed the concept of Semantic Web in 2001.But with further research, more and more researchers noted the gap between today's network and the Semantic Web. On one hand, a large number of HTML document in current World Wide Web is prepared for human reading, not for machine processing. Web knowledge marked by natural language and HTML that the computer can not use the semantic information knowledge, can not meet the high demands of users. On the other hand, Semantic Web required network resources added semantic knowledge to the document. The machine can handle the document. Information extraction, word segmentation, POS tagging, syntactic analysis, semantic annotation technologies are relatively mature, the semantic annotation of network resources more and more attract people's attention, and is becoming an important part of Semantic Web study.Ontology through define the concepts and the relationship betweem the concepts, determine the precise meaning of concept, provide a commom understanding of konwledge in this area, realize knowledge sharing and reusing, achieve interoperability between different systems and models of this area. Ontology as a knowledge can provide knowledge sharing and reusing in knowledge level, providing a shared conceptualization model of knowledge for Semantic Web.Network resources semantic annotation is the appliation of annotation thinking and technology for corpus. The second chapter describes framework semantics, CFN engineering and Chinese Framework ontology construction. The third chapter focuses on describing the proess of network annotation, and the implementation of these processes. Introduce CFN ontology construction and the related knowledge,background of network annotation, analysis network annotation process and realization process, proposed the model of network annotation based on CFN ontology. Elaborate separately from the ontology management, information access, dependency treatment, the role labeling and the results stored of network annotation. The forth chapter, see the text-matching and maximum entropy as two case explained the algorithm and implementation of role annotation. And give an example of "steal".