Research On Recommendation Algorithm Based On Relative Preference And Curiosity Model

With the advent of the Internet age,various Internet companies have become more and more connected with people's lives.The recommendation system is applied in more and more fields to save the user's time cost and help the user to mine their potential interest so as to improve the user's product experience.The recommendation system is currently an inseparable part of the Internet products.Excellent companies and scholars in various countries are continuously researching and focusing on the development of the recommendation system.This paper proposes a solution method based on quadratic optimization method aiming at inaccurate positioning of user interest in traditional recommendation technology and the popularity and simplification of recommendation results.The aim is to solve the problem of novelty of recommendation results.The specific research contents are as follows.First of all,aiming at the problem that the traditional recommendation system user's interest positioning is inaccurate,a user preference modeling method based on relative preference is proposed in the user modeling phase of the secondary optimization method.Through the elimination of rating biases for all users in the scoring matrix,the evaluation criteria of all users are unified.Based on this,using the relative preference idea,the target user's preference of the target project and the user group's preference for the project are compared and analyzed to determine the unique interest preference of the user.Second,in order to solve the problem of poor novelty of the traditional recommendation algorithm,a second optimization recommendation algorithm based on curiosity model is proposed.The curiosity model in psychology is introduced into the recommendation algorithm,and the user's curiosity curve is predicted and modeled by quantifying the historical stimulus received by the user,so as to judge whether the target item is moderate to the user's stimulus,and thus The recommendation list is screened to improve the novelty of the recommendation results.Finally,the recommendation algorithm based on relative preference and curiosity model presented in this study is compared with two other representative recommendation algorithms,and the experimental results are performed through the three evaluation indexes of accuracy rate,recall rate,and average popularity.analysis.