Research On Web Page Ranking Learning Based On Generalized Regression Neural Network

With the rapid development of the Internet and the number of data such as blogs and articles published by users through the Internet has grown exponentially,the era of big data has arrived.For example,Google,Baidu,Yahoo,and other web search engines contain data in the terabytes and even the PB level.In real life,the main task of the user every day is to retrieve the required data from the Internet,and the task of web search engine is to quickly find the data resources that satisfy the user from such a large data set and sort them according to the relevance of the retrieval conditions to return to the user.Learning to rank plays a vital part in search engines.How to quickly retrieve data from such a large data set and sort by relevance of retrieval conditions is an important research topic in the learning to rank model.The traditional web learning to rank model faces the shortcomings of long training time and low accuracy when performing information retrieval.This paper combines the characteristics of general regression neural network and proposes web page learning to rank model based on general regression neural network.Firstly,this paper introduces general regression neural network to establish a web page model of learning to rank.General Regression Neural Network is an improved model based on Radial Basis Function Network.General Regression Neural Network has the advantages of fast learning speed,few training samples,high real-time performance and convergence to the global optimal solution,which can improve the long training time of traditional learning to rank models.Secondly,this paper uses the genetic algorithm to search for the optimal value of the smoothing factor parameters,thus improving the low accuracy of learning to rank model.Thirdly,General regression neural network has the disadvantage of high spatial complexity.This paper proposes to use the technique of blocking to improve.The locality principle of the computer is used to solve the problem of excessive spatial complexity of the generalized regression neural network.Finally,the experimental analysis shows that the learning to rank model based on generalregression neural network has higher query effect.