Study On Technique Of Dynamically Modeling Web Pages Base On Web User's Search Intentions

When a user issues a keyword to search engines for query, her search intention may be three classes:(1) Reach a particular website (Navigational Query); (2) Find the information assumed to be available on the Web in a static form (Informational Query); (3) Reach a website where further interaction will happen (Resource Query). Though most search engines can provide some functions for interacting with users, it is still difficult for the user expresses her intention accurately. Moreover, it is not purpose for search engines to obtain users' intention, but the real purpose is to provide different information for different users with different intentions. However, the results returned by current engines often contain both the desired information and the unwanted information to the users. Especially, not all results required by the user can be ranked in the first some pages.In this thesis, users' search intentions and the relevant information requirements of the intentions are generalized, and put forward that the relationship among the three intentions is that Informational Query is the initiation of Navigational Query and Transactional Query, or Navigational Query and Resource Query are the goal of Informational Query. Based on this proposal, the idea of dynamically modeling web pages based on the search intentions is proposed. According to the idea, a simple and easy algorithm of clustering web page is designed and implemented. In the algorithm, the similarity may be containing keyword, containing type, and containing format, respectively, so that the propose of clustering web pages returned by search engines can be implemented according to above three intentions. Moreover, different results can be returned for different intentions.Comparing with related work, our work does not focuses on the obtaining users'intention and classifying users' intention, but based on the classification of users'intention proposed by the related work, models the web pages of results returned for users'further selection. The method can not only dynamically model web pages based on users'search intentions and provide the retrieval navigation for the users with different intentions based on the web page model, but also show more relevant results returned to the users'intention.