Study On Tag Ambiguity And Redundancy In Social Tag Recommendation

The tag recommendation algorithm of Folksonomy tag recommendation system does not consider the trouble of Ambiguity and redundancy on tag, these tags reduce the system's recommended quality and hamper the effort to judge recommendations as well in Folksonomy, if we apply Ambiguity and redundancy on tag to the tag recommendation algorithm which solve the above problems, improve the recommended quality, and provide a more reasonable evaluation.The tag recommendation of Folksonomy tag recommendation system does not consider the trouble of Ambiguity and redundancy on tag, reduce the system's recommended quality. The recommended quality of tag recommendation system significantly improved when we applied Ambiguity and redundancy on tag to recommendation algorithm. In addition, when we add tag ambiguity and redundancy processing, the accuracy of KNNut tag recommendation algorithm to achieve significant improvement, This is because dealing with ambiguity tag increase the accuracy of user modeling; but the recall of POPr tag recommendation algorithm to achieve significant improvement, This is because the algorithm recommended more redundant tag.The clustering of tag and resourse can affect the accuracy of extracted ambiguity tag and redundant tag, If clustering is not good, will lead to overestimation of the tag ambiguity,and reduce the accuracy of extracted redundant tag. So we uses Kernel Clustering algorithm to deal with the problem of Ambiguity and redundancy on tags, this approach improve the effect of clustering because it can identify, extract and enlarge useful features of the sample by non-linear mapping. Our experimental results show that:Kernel Clustering a algorithm has better results in the clustering of tag and resource, the extraction of Ambiguity tag and redundancy tag are more accurate.,and the recommended quality is more higher.