Research On A Content Analysis Method Used For Ontology Ranking

As the development of the semantic Web, ontologies technology is paid more and more attention by researchers. It becomes a very important research subject that how to help users get the useful ontologies and reuse them effectively. In this thesis, it mainly focuses on ontologies ranking which is the key issue in the ontologies search area.First, it introduces the background of the subject including the summary about the semantic Web and ontologies. Then it classifies the ranking problem in the World Wide Web according to the ranking objects. To every category, it analyses the main method and takes some typical algorithms for example.Second, in order to help users to understand related ontologies, it proposes an algorithm of extracting the ontology schema. In this algorithm, it gets the ontology concept model according to the structure of the ontologies. Based on the ontology concept model, it obtains the ontology key words set and assign the weight to every concept. Then, it gets the important properties set according to the ontology concept model and the weight of the concept. The ontology schema is gotten combining the ontology topic set with the important properties set.Third, aiming at the problem that the ontology ranking result is not precise enough using traditional method, it proposed a new content analysis method. It gets the topic relatedness according to the match between the keyword and the weighted concept and gets the context relatedness by analyzing the context of the key word. And then the ranking score can be calculated by combining the topic relatedness and context relatedness.Finally, according to the research above, it designs and implements the content analysis method used for ontology ranking. It gets the test set by using the ontology search engine Swoogle to test the precision of this method. The experimental result is evaluated by the Pearson Correlation Coefficient between the subjective result and the experimental result. The experimental result shows that comparing to the OntoRank algorithm, the result gets from this method is more similar to the subjective ranking result.