Context-aware user modeling method and sequence recommendation method based on meta-path

Abstract

The invention belongs to the field of user modeling and recommendation systems, and specifically relates to a meta-path-based context-aware user modeling method and a sequence recommendation method, aiming to solve the problem that the existing user modeling methods only consider the simple sequence relationship of user historical behavior, Complex user preferences cannot be captured; the accuracy of the existing recommendation system recommendations is not high, and the recommended content does not have diversity. The present invention acquires local interest representations based on the sequence of items accessed by users, obtains context-aware global interest representations by modeling heterogeneous networks, and combines local interest representations and global interest representations to obtain user comprehensive interest representations. Based on the user's comprehensive interest representation and the items to be predicted, the recommendation results are obtained through the recommendation model. The user modeling method of the present invention can more comprehensively capture user interests, and is beneficial to improving the accuracy of the recommendation system; the meta-path-based context-aware sequence recommendation method has high recommendation accuracy and more diverse recommendation contents.

Description

technical field [0001] The invention belongs to the field of user modeling and recommendation systems, and in particular relates to a meta-path-based context-aware user modeling method and a sequence recommendation method. Background technique [0002] On social media networks, users contribute a large amount of behavioral data for user modeling and personalized services, such as personalized recommendations on e-commerce platforms, news recommendations on information platforms, and so on. Recommender systems play an increasingly important role in matching user needs with rich items from various online platforms. In building an effective recommender system, a key factor is the ability to model user interests to accurately characterize and understand users' dynamic and changing tastes and interests. [0003] Traditional recommendation algorithms, such as content-based recommendation methods and collaborative filtering recommendation methods, usually consider the user as a st...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above problems in the prior art, that is, in order to solve the existing user modeling method, only consider the simple sequence relationship of user historical behavior, and cannot capture complex user preferences; the accuracy of the existing recommendation system recommendation is not high, and the recommended The content does not have diversity, the first aspect of the present invention proposes a meta-path-based context-aware user modeling method, the method includes:

Images