Improvement And Implementation Of A Recommending Algorithm Based On Scoring Space In Trust Relationship Network

Since people stepped into the 21 st century,the Internet and information technology have been growing explosively.The network is full of variety information.With the information amount heavily growing,it has brought unprecedented convenience as well as unprecedented distress to people.On the one hand,the way people get information becomes very convenient,once you connect to the internet you can find any information you desired;on the other hand,the time one has is limited,the information one can handle is also limited.Thus there is a contradiction between the increasing amount of information and the unchanged information processing capacity.Then recommending system comes out.Its purpose is to make the information presented have a higher relativity with users through technical means,the higher the better.In recent years,with the integration of social network and people's daily life,recommending system based on social network has become a hot research field.In this field,the trust relationships among users have attracted much attention.The results of scientific research show that friends have far-reaching effects on human behavior compared to family members,and the trust relationships between friends play an important role in improving the performance of recommending system.In recent years,researches about trust have emerged lots of outstanding papers,and researches about distrust are also becoming more and more.But in most existing reasearches,trust and distrust are manuly seperated.This caused some problems.The main work of this paper is summarized as follows:1.This paper analyzes the current trust relationship measurement,then we try to dig out the possible hidden problems with real examples,with these reasons we introduce another trust relationship measurement-scoring space,and solve the provenance tracking problems in the propagation with traditional trust relationship measurement by the way.2.Considering the hidden information in the topology of the new trust relationship network,this paper proposes a new improved trust aggregation algorithm considering the decaying problem in long distance propagation and the authority degree problem in the expansion of the trust relationship network.3.Considering the characteristics of data presented by scoring space and that the recommendation by slope-one is too popular,this paper proposes an improved score prediction algorithm,we reduce the range of the weighted data to the trusted neighbors and make use of the scoring space data to calculate the weighted result.4.Based on the new algorithms proposed in section 3 and 4,this paper designs an experiment to verify the performace of the new algorithm proposed and finally draw the conclusion with the results of the experiment.