An Ontology-based Personal Recommendation System

With the rapid development of information technology, Intelligent Turing System,as an measure of aided instruction, has becoming a research focus of many scholars inthe E-Learning field. But most of them lack effective recommendation mechanism oflearning resources, which makes system can not accurately recommend resources forstudents facing massive resources. As a result not only will students’ interest in learningdecline, but also the system can not give full play to aided instruction.On the basis of relevant theories, an algorithm of personalized recommendation, inwhich ontology is introduced to represent knowledge, is proposed in order to solve theproblem of recommending learning resources from massive items bank for students.This research can avoid unexcepted recommendation caused by ignoring information ofresources and provide a new thought of ITS’s application. The details of the research aredescribed as follows:First, this thesis researches the theories of ontology, semantic relevancy and perso-nalized recommendation, including ontology’s construction, traditional caculation me-thod of semantic relevancy, algorithms of personalized recommendation and so on. Thekey point is introducing the technologies used in this thesis, the theories and caculationmethod of Analytic Hierarchy Process, which is the classical decision algorithm.Second, an algorithm of ontology-based semantic relevancy, which uses the Ana-lytic Hierarchy Process to integrate four semantic relevancy’s indicators: semantic dis-tance, semantic contact ratio, concept hierarchy difference and semantic weight into acomprehensive metric value to evaluate the semantic relevancy between concepts of aontology. Comparative experiments, which comparative the data generated by this algo-rithm and similar others to evaluate the semantic relevancy on different ontologies, aredesigned to demonstrate that the proposed algorithm can effectively calculate the on-tology-based semantic relevancy and the results are more in line with subjective judg-ment than others.Third, an algorithm of ontology-based recommendation is proposed. By dividingthe recommendation into four steps: ontology-based concepts extension, resources query based on semantic set, caculation of semantic relevancy and caculation of rec-ommendation degree, one concept is changed into a set of resources which has a certainsemantic contaction with it. And by grouping concepts into different semantic sets andcaculating the semantic relevancy between them, the recommendation degree of the re-sources can be get. At the same time, the experiment data proves that the recommendedresults meet the subjective expectation and the algorithm has a significant recommend-ing effect.Last, an application system, namely an intelligent turing system, based on two al-gorithms mentioned above is proposed and designs three modules: recommendation ofpreview, recommendation of synchronization and recommendation of review, whichproves the theory and application value of this algorithm of ontology-based recommen-dation.