User confidence coefficient and time context-based collaborative filtering recommendation method

Abstract

The invention discloses a user confidence coefficient and time context-based collaborative filtering recommendation method, and relates to calculation of similarities between users in the collaborative filtering recommendation method. Aiming at the deficiency problem of traditional similarity measurement methods and dynamic change conditions of user interests under the condition that user score data is extremely sparse, the method discloses the user confidence coefficient and time context-based collaborative filtering recommendation method. The method comprises the following steps of: constructing a scoring matrix between users and projects; calculating similarities between the users through the user confidence coefficient and time context-based collaborative filtering recommendation method; selecting an optimum adjacent user set according to a sorting result of the similarities between the users, or setting a similarity threshold and selecting the users exceeding the threshold value as neighbors of a target user; and after obtaining nearest neighbor set of the target user, taking the similarities as weights to obtain prediction, for unscored projects, of the target user, so as toform a Top-N list and recommending the Top-N list to the users.

Description

technical field [0001] The invention belongs to the technical field of personalized recommendation methods, in particular to a collaborative filtering recommendation method based on user confidence and time context. Background technique [0002] Today, resources on the web are exploding. On the one hand, people obtain more and more abundant information through the Internet, which brings great convenience to life; on the other hand, while the massive information space brings users more diversified choices, it also makes users lost in the in the ocean. Although traditional search engines can alleviate users' information retrieval needs to a certain extent, they present the same sorting results to all users, and cannot proactively provide personalized services for different users' hobbies. In this context, personalized recommendation came into being. [0003] The essence of personalization is to provide different services for different users. In real life, the role of a per...

AI Technical Summary

Problems solved by technology

[0004] But at present, the personalized recommendation system has low accuracy and reliability

Many personalized recommendation systems cannot fully make recommendations according to the changes in users' interests, especially the changes in users' interests in different periods of time, making the recommendations lose their accuracy and pertinence.

Images