Multilingual Tags Clustering And Its Application

Tags are products of Web2.0,they are widely used because of their personalization features. Internet users always use tags to describe documents, pictures and videos. However, information retrieval becomes more and more difficult with the large expansion of tags. On the other hand, semantic heterogeneity of tags is increasing, clustering tags can solve this problem, and we can discover social networks based on clustering tags.In this paper, tags from blogs are extracted by using different algorithms respectively. We also optimize the extraction tags and compare them with users’tags. Experiment results show that the performance of these two algorithms is very closely.Divisive hierarchical clustering algorithm is used to cluster tags. Users’tags and extraction tags are clustered respectively. The author also tries to discover social networks by clustering tags. Experiment results show that the performance of users’tags is better than extraction tags.The author realizes tags mapping based on machine translation. Multilingual social networks are found based on this, and the author also realizes recommendation of friends. Experiment result shows that multilingual social networks are more complex than monolingual social networks.Lastly, the author studies tags application. In this paper, she uses microblog to monitor product information. Firstly, she collects related users’tags and microblog from a microblog system. Then, the tags and microblog are clustered. According to the clustering results, the social networks and hot topics about the product can be generated. The product name is used as query on a famous microblog system. Experiment results show that the microblog mining is applicable for monitoring of product information.