The Application Of GA And Relevance Feedback In Query Optimization

Users usually retrieve the information by inputing some keywords into the retrieving system.The keywords which are inputed by users for the first time are usually not exact enough , the information reflected by these keywords is also limited,and the users'fimilarity of the information retrieving system is different,these reasons makes that the users can't express their query intention exactly. Aiming at this actuality,numbers of scholars developed the research on query optimization techniques.Because of GA's(Genetic Algorithm) robustness,adaption and powerful searching ability and is applied into the query optimization widely;Relevance feedback is another technique that used usually by query optimization and it has demonstrated its excellent performance in the domain of query optimization.Therefore,we adopt the Vector Space Model as the our retrieving model ,and we make our first attempt to optimize the query only using the genetic algorithm ,and get the optimized query directly.The attempt achieves the obvious effect in the end ;Then combining relevance feedback technique and adjusting the GA's function, we improve a query optimization method based on GA and relevance feedback.The method can improve the query optimization effect.The main content and the main fruit of our thesisr are the following:Firstly, some research was done on the GA's function in query optimization.Based on the analysis of current typical application of GA-based query optimization, by ameliorating the fitness function and the genetic operators,we designed a new GA-based query optimization method to seek the optimized query vector.The new method can improve the query optimization's effect.Secondly,we probe into the effect of the combination of GA and relevance feedback on query optimization.By analyzing the traditional relevance feedback technique's effect on query optimization and combining it with GA subtly,we propose a new query optimization method based on GA and relevance feedback.We introduce a new concept of query adjusting vector and utilize the above GA to acquire the query revising vector .We combine the initial query and relevance feedback to form the relatively optimized query. This method improve the query optimization's effect further.Finally,we validate our methods by experimenting widely.We make use of the five famous standard test collections(Cranfield,Medline,CISI,NPL,CACM)in the world to...