Research On Content-based Recommendation Based On Semantic

The advent of the Internet has changed the way people get and share information, and thebirth of Web search engines is an important milestone in the history of the network. After1998,the link analysis emerged in the context of search and took the market rapidly. Google Inc. hasgrown, in less than10years, from a startup to a dominant player in the technology sector.However, as the demands of the Internet increases, personalized recommendation system asthe representative of the new generation of Intelligent Web had got more and more attentionof researchers. Among the recommendation strategy, content-based recommendationalgorithm as an item-center recommendation strategy can solve the collaborative filteringrecommendation’s cold start problem. But the problems of obscure eigenvalue and featuresmismatch in traditional content-based recommendation system make it difficult to accuratepositioning of merchandise attributes.Semantic Web, which initiated by the father of the Internet Tim Berners-Lee, provides agood knowledge of the organization, reasoning and retrieval way to get commodities attributes.Thus, this dissertation used the theory of semantic web, and described a new content-basedrecommendation system based on Google similarity distance. To extract the keywords fromproducts as tags, then by using Google similarity distance get the relationship between tags andmerchandises, and building merchandise’s ontology Using Internet resources to build the taghierarchy, by reasoning on ontology get more implicit information, and finally through the tagsas a bridge to find similarity between the merchandises, based on this to recommend othersimilar items. On this basis, the dissertation gave an experiment to prove the effectiveness ofthis method.