Research And Application On Optimizing Commercial Recommendation Algorithms Based On SVD And Web Spammer Cluster

Recommendation algorithms is widely used for various kinds of website in modern society.The developer of website hope to use recommendation algorithms to improve user experience.User experience is of great importance for websites which have same product.While in modern society,web spammers have challenged the accuracy of recommendation algorithms.For recommendation algorithms use all data for prediction,web spammer may disturb recommendation result.This paper hopes to put forward a new recommendation algorithm for large-scale spammer in order to avoid the disturbance of spammer.After read a large number of literatures,this thesis introduces the concept of recommendation algorithms and research background,then introduce some classical recommendation algorithms,delves into the part of the recommendation algorithms and process,and make summary of these recommendation algorithms.After this,the paper finish the following works.(1)put forward SVD optimizing commercial recommendation algorithms Based on cluster detection of web spammer,divide it into two parts,spammer dection and rating matrix prediction.(2)research the behavior pattern of spammer.Put forward spammer dection algorithms based on the pattern and rating data.(3)research classical recommendation algorithms,put forward recommendation algorithms based on the requirement of spammer detcion.(4)check input and output,analyse algorithm feasibility,finish implementation on the algorithm using data set and database.(5)analyze the accuracy of SVD optimizing commercial recommendation algorithms Based on cluster detection of web spammer,analyze the accuracy of other method and compare them.