A New Kind of Personalized Search Method Based on User Transaction Records
Xu Ka · Computer Knowledge and Technology · 2015
In this paper, we proposed a new method which is used in B2B search results ranking. We could implement personalized search results ranking based on the user transaction records. We using user transaction records to build trading directed graph, building transfer matrix based on trading directed graph, and calculating the credibility of users by using the thought of Pagerank~1. We could get preliminary search results according to the keywords matching,And we ranking the results one the basis of the credibility of users and the user's preference. In addition we add Feedback coefficient and the correction coefficient into transfer matrix, which could effectively prevent the behavior that doing malicious trades to get high credibility.