A group recommendation method based on combination learning

Abstract

The invention provides an ensemble learning-based group recommendation method comprising: manner 1, acquiring personal recommendation lists of all users based on the K-nearest neighbor algorithm, sorting out recommendation lists of members of a group A, and obtaining a recommendation list of the group A from the recommendation lists of the members of the group A according to the ensemble learning method; manner 2, accumulating grade lists of the members in the group A, calculating the average value to obtain a grade list of the group A, and, based on the K-nearest neighbor algorithm, generating a recommendation list of the group A according to grade lists of members not in the group A; manner 3, establishing a preference model of a group, expressing the preference of each user with a feature factor vector, obtaining a feature factor vector representing group preference according to the preference vectors of the individual users and generating a recommendation list of the group A according thereto; obtaining a more accurate recommendation list from the three recommendation lists for the group A generated via the above-mentioned three manners based on the ensemble learning method.

Description

technical field [0001] The present invention relates to the field of computer technology, and more specifically, to a group recommendation method based on combination learning. Background technique [0002] Nowadays, human beings are in the era of information explosion. In order to effectively analyze and utilize these data, recommendation system is widely used as an advanced business intelligence platform to help e-commerce websites provide decision support and information services to their customers. [0003] With the rapid development of social networks, more and more people are willing to share their group activities, such as having dinner with colleagues, watching movies with spouses, etc. This motivates the study of group recommendation aimed at making recommendations for a group of users. [0004] The group recommendation problem is very challenging, because different users in the group have different preferences, and how to make recommendations to maximize the aver...

AI Technical Summary

Problems solved by technology

The main disadvantage of the memory-based method is that it ignores the interaction of members in the group

The model-based method uses some probability models to simulate the process of group recommendation. The main disadvantage of this method is that some models may not exist in reality.

Images