Cross-domain recommendation method based on multi-view knowledge representation

Abstract

The invention provides a cross-domain recommendation method based on multi-view knowledge representation. The method comprises the steps of integrating different projects in a heterogeneous graph formaccording to similar attributes of the projects in different fields to form a plurality of views, taking the views as inputs of a graph attention network respectively, and obtaining initial knowledgerepresentation of the projects under the views through the graph attention network; taking the initial knowledge representation of the project under each view as the input of a multi-head attention network, obtaining and integrating project representation vectors with user preferences under different views through the multi-head attention network, and obtaining the final representation of the project with the user preferences; and recommending a corresponding project in the target domain to the user according to the final representation of the project with the user preference and the information of the target domain. The multi-view multi-head attention network learning method is set among multiple fields, project knowledge representation is fully learned, cross-field recommendation is carried out, and therefore the recommendation effect of the target field is improved.

Description

technical field [0001] The invention relates to the field of big data technology, in particular to a cross-domain recommendation method based on multi-view knowledge representation. Background technique [0002] Due to the exponential growth of the amount of information available on the web, recommender systems have become inevitable in our daily lives. However, when users have less feedback on items or new users and items are added to the system due to lack of historical information, the recommendation performance will be extremely degraded, which is the so-called single-domain data sparsity problem and cold start problem. [0003] A natural solution to the above-mentioned problems in single-domain recommendation is to use data information from other domains to enrich user models to generate better recommendations, that is, cross-domain recommendations. For example, the genres of movies a user likes can be derived from the genres of books he likes. Furthermore, cross-doma...

AI Technical Summary

Problems solved by technology

[0014] The disadvantages of the above-mentioned cross-domain recommendation schemes in the prior art are: most of them directly apply the collaborative filtering algorithm to cross-domain scoring data, or apply transfer learning to cross-domain recommendation, or use knowledge aggregation, which involves The knowledge mostly adopts the scoring matrix between users and items

However, in addition to the directly expressed rating data between user items, there are also fine-grained attribute preference associations between users and items. The existing models do not fully consider how to describe the fine-grained attribute characteristics and attributes of items in multiple fields. Relationships between items to learn item representations

[0015] Heterogeneous Information Networks (HIN) can represent directed graphs with multiple types of nodes and edges and various complex relationships. Most of the existing heterogeneous graph learning methods are non-end-to-end learning methods, which rely on The cumbersome process requires the use of random walk methods to manually select meta-paths or meta-graphs, or to perform knowledge representation construction, similarity, or off-line calculation of exchange matrices in heterogeneous graphs. inefficient

[0016] In the existing heterogeneous graph algorithm, the influence of different item attributes (views) in the heterogeneous graph data on the user's preference is not distinguished, but in reality, for different movies, users like to decide what to watch according to different movie attributes. Content, such as director, actor, and genre, have different influences on whether a user watches a movie

Images