Research On The Method Of Automatic Semantic Annotation Based On Ontology

Along with the increase of information content on the Internet,it's more difficult for people to quickly and accurately obtain the required information.The traditional method is to carry out artificial semantic annotation of web information,however,the manual annotation is clearly not feasible on the existing massive web.Therefore,how to improve the quality of automatic semantic annotation is a focus of research on the dissertation.Ontology can better describe concept the information in terms of semantic and knowledge,so the ontology is introduced into the traditional semantic annotation method.Through the analysis and discussion of the semantic annotation method,it is understood that the natural language information extraction and annotation algorithm is the key to improve the semantic annotation.Delved into natural language relation extraction method and semantic annotation algorithm,this dissertation mainly studied on how to efficiently extract ontology knowledge based building semantic relation between the concept of attributes,and study how to improve the accuracy of marked results by the improved semantic annotation algorithm.In this dissertation,the main work done is as follows:(1)A new relation extraction method is proposed to solve the problem of relation extraction from open Chinese free texts.In order to alleviate the difficult problem of relation triples extraction,put forward a method is based on the relationship between attribute and concept instance triples,extracted a large number of instances of concept and relation triples includes not only explicit relation triples also contains an implicit relation triples.On the basis of this,the relationship triple construction contains noise and error,in view of the relationship between the ternary group contains noise and wrong question,using Adaboost based iterative algorithm of collaborative training methods to strengthen the relationship between extraction model.Experiment is carried out on the text of the encyclopedia entries in the field of university,and the experimental results show that the method can obtain better performance.(2)In the process of semantic annotation,in order to eliminate the ambiguity problem of the text in a given named entity and the mapping of the knowledge base entities.A context based semantic similarity value of the sorted named entitydisambiguation method is put forward.Disambiguation method included three sections that entity preprocessing,constructing candidate list of entities and similarity value ranking algorithms.In view of the problem of the named entity reference multiplicity,the new entity was used to represent the preprocess method to extract the standard entity.Then used the online encyclopedia in Chinese to construct the semantic knowledge base,and got the semantic list of standard entities.At the same time,this paper also put forward using the similarity value ranking method for solving standard substance and semantic list mapping referential ambiguity problem,for in the knowledge base not found semantic entity disambiguation processing by clustering algorithm.The results of the experiment show that the proposed method can effectively reflect the real data set of Chinese web pages to the corresponding non-ambiguous entities in the knowledge base.