A Recommendation Method Fused with Text Semantic Vectors and Neural Collaborative Filtering

Abstract

The invention discloses a recommendation method that combines text semantic vectors and neural collaborative filtering, including: a data preprocessing module acquires user comment text and item metadata; a user comment characterization module generates user comment embedding vectors based on user comment text; item content The characterization module generates the embedding vector of the item content according to the item description text. The recommendation model inputs the embedding vector of user comments, the embedding vector of item content, and the one-hot encoding of user ID and item ID into the hybrid recommendation module and rating prediction module in sequence to predict user ratings. The present invention introduces a text paragraph vector embedding representation method to realize the text representation learning of user comments and item content, and input the obtained embedding vectors into the user emotion analysis network and the item content analysis network respectively, and the output is regarded as the user and item content Collaborative attention, separately acting on user-item interaction sequence modeling, improves the score prediction performance of recommendation models.

Description

A Recommendation Method Fusing Text Semantic Vectors and Neural Collaborative Filtering technical field The present invention relates to the technical field of recommendation, be specifically related to a kind of recommendation of fusion text semantic vector and neural collaborative filtering recommended method. Background technique As an effective tool to solve information overload, recommender systems have received more and more attention in both academia and industry. Note. Recommender systems, as a technique that attempts to predict user ratings or preferences, generate and Offer advice or recommend items. Collaborative filtering method is one of the techniques commonly used in recommender systems, which is based on the following assumptions: Items that households liked in the past will also be liked in the future. This technology uses only explicit user rating information to generate recommendations, And often suffer from cold starts, scalability and sparsi...

AI Technical Summary

Problems solved by technology

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

Images