Research On Social Tagging Recommendation Based On Data Warehouse And The Analysis Of Semantic

With the rapid development of Web2.0technology and electronic commerce，theamount of data in the network is exploding，which leads to the problem of informationoverloading on the network，The user needs to pay a high price to resources to findinteresting information from the sea of the data. In order to reduce the cost, therecommendation technology based on social tag has appeared.Social label is a keyword or phrase in the tagging resources users independent choice,social tag can not only help users better organization, management of interest information,but also can find user preferences referring users to meet their interest preferences ofresources according to user label information. So in this background，how to accurately andeffectively recommend tags, and discovery of user’s preferences and recommend resourcesto user according to the record of user’s tag, which become an important direction of currentresearch.At present,there are many mainstream tag recommendation methods, but in theprocess of recommendation on the relationship of three elements user, resource and tag isgenerally considered insufficient, but the multidimensional data of the warehouse provides away can well reflect the relationship of the three elements, So this paper researches how toapply the technology of data warehouse to tag recommendation and efficient recommendedto the user; at the same time, because the three elements in the data warehouse of semanticrelations cannot be well represented, this paper also proposed the model of weighted tuplessemantic based on social tag. The following is the main work of this paper:1. The association relationship of users, resources and tags is inadequate considered ontag recommendation. This paper proposed a model of social tag based on data warehousetechnology. The model uses a multidimensional of data warehouse technology and theconstruction multidimensional data set of tag, creating the data mining model and theunderlying multidimensional data model, which uses OLAP analysis the data andMicrosoft association rules of data warehouse to min multidimensional tag data set rules.Then the model achieve social tag recommendation through using data warehousetechnology, resolving the problem of low recommendation precision which is lack ofconsidering of the relationship of user, resource and the tag.2. Using data warehouse technology based on recommended tag can not well reflectthe semantic relationship of users, resources and tags. This paper proposed a recommendation model based on the latent semantic analysis weighted tuples. The modelintroduces the method of the social network structure analysis add weight to the tags tuple,create a weighted3-Dim tensor model and use HOSVD to process this model. Then throughthe latent semantic analysis of tuple, getting the weighted tuples set that can reflect theuser’s interest and generate the result set of recommendation. Using this model to resolvethe problem of low accuracy and efficiency, which is lack of considering the semanticrelationship of users, resources and tags.Through the tag data set of the typical tag website Delicious, the paper has validatedthe social tag recommendation model based on data warehouse technology and the socialtag recommendation model based on the weighted tuples latent semantic analysis.