Extracting Cluster Tags For Search Results Clustering

With the increasing expansion of internet information, the proliferation of information is becoming worse. All kinds of search engines, as the primary means of access to information, have to help users quickly locate a particular knowledge acquisition. How to guide users to relevant information or push related information to the user, have become pressing issues of search engines. Search results clustering will help search engine solve the solution to this problem, providing the user information of behavior guidance, data distribution. Search results clustering will provide a present form of search results, and it will help with search ranking, related search. It has very broad application prospects with text retrieval, digital library management, entity relationship mining.Organizing search results of search engines into clusters facilitates users'quick browsing through search results. The traditional clustering methods are inadequate since they can't act fast enough and they do not generate clusters with highly readable tags.In consideration of traditional methods and rapid response to the needs of search engine, we take the clustering problem as a ranking problem. And hierarchical clustering will be used to cluser the documents and tags will be extracted. This study covers the following content:1. The tag extraction strategy. With the help of all kinds of semantic information, we will extract tags with good quality as far as possible.2. Take ranking model to generate cluster candidates.3. We introduce an incremental hierarchical clustering algorithm in order to do clustering.4. With comprehensive utilization of results information and user behavior information, we propose a method to guide the label extraction.These methods are all verified through experiments.