| At this time of information explosion, it has already become a problem that information is toomuch and mess when people want find answers from the internet.As a query tool,the searching enginehas been applied more and more widely with the popularity of Internet.However, as the techniquewhich most of the search engine use is dependent on the keyword-based search, there often returns alot of useless information, resulting in low query efficiency.In recent years, how to improve theprecision and recall of query expansion through the association rules mining, has become a hot anddifficult research problem.The association rule mining is the most important and the most basic functions in datamining.Therefore, how to find more keywords through association rule mining to improve theefficiency is one of the main research directions on query expansion. This thesis analyzes theassociation rule mining and query expansion research background, research status, and thenintroduces the relevant basic theory, including the theory of association rule mining and related keytechnologies, highlighting the various association rule mining algorithm, and the FP-growth algorithmfor the analysis shows. After all this, we explore the query expansion technique and related researchin-depth.Based on the above discussion, this thesis is proposed to improve the following two points:firstly, we improve the way to find frequent patterns, whch proved to reduce the time complexity.Secondly, after the step of association rule mining, there will be a heavier burden on the queryexpansion.So we proposed a model based on web-tagged information, through the words on the pageto mark the location information and the right weight, to quantify the web pages.Then we can findmore relative pages for the next step of association rule mining, which could also ensure the recallrate as much as possible on the basis of high correlation.These two points have been experimentallyproven. |
PDF Full Text Request