Recommendation method integrated with text semantic vectors and neural collaborative filtering

Abstract

The invention discloses a recommendation method integrated with text semantic vectors and neural collaborative filtering. The recommendation method comprises the steps that a data preprocessing moduleacquires user comment texts and article metadata; a user comment representation module generates an embedded vector of a user comment according to the user comment text; and an article content characterization module generates an embedded vector of the article content according to the article description text; a recommendation model inputs the embedded vector of the user comment, the embedded vector of the article content, the one-hot codes of the user ID and the article ID into a hybrid recommendation module and a score prediction module in sequence to perform user score prediction. A paragraph vector embedding representation method of the text is introduced, representation learning of the text of user comments and article contents is realized, the obtained embedding vectors are input into a user emotion analysis network and an article content analysis network, and the output of the embedding vectors is regarded as the collaborative attention of the user and the article; the invention is applied to interaction sequence modeling of user articles, and improve the score prediction effect of the recommendation model.

Description

technical field [0001] The invention relates to the technical field of recommendation, in particular to a recommendation method integrating text semantic vectors and neural collaborative filtering. Background technique [0002] As an effective tool to address information overload, recommender systems have received increasing attention in both academia and industry. A recommender system, as a technique that attempts to predict user ratings or preferences, generates and provides suggestions or recommended items for users by employing various strategies. Collaborative filtering is one of the commonly used techniques in recommender systems, which is based on the assumption that items that users liked in the past will also like in the future. This technique only uses explicit user rating information to generate recommendations, and often suffers from cold-start, scalability, and sparsity issues. [0003] Matrix decomposition is a key technology in collaborative filtering algori...

AI Technical Summary

Problems solved by technology

However, there are some tricky issues when building a hybrid recommendation model

Images