Research On Recommendation Algorithm Of Recommendation Topic Analysis Based On Deep Learning

In recent years,with the popularity of the Internet and the advancement of computer technology,the amount of information has grown exponentially.People enjoy the convenience brought by the Internet,but also suffer from "fragmentation of information" and "information overload." The personalized recommendation system can not only meet the individual needs of users,improve the user experience,but also greatly improve the user's behavior conversion rate under specific business scenarios.It has become the mainstream method to solve the problem of "information overload".At the same time,with the rapid development of artificial intelligence,deep learning is increasingly applied to the recommended field.On most of today's e-commerce platforms,writing reviews of products is not only a core feature,but also a very historic event.Rich reviews are especially powerful in capturing user preferences and portraying product features,as it contains a lot of information.This thesis mainly studies the recommendation topic recommendation algorithm based on deep learning and implements an online real-time recommendation system.Combine deep learning with recommendation algorithms with comments.The research contents are as follows:(1)An attention neural network model with a screening mechanism was proposed.First,the network first performs one-hot encoding in the embedded layer,calculates the user(commodity)comment list through the same matrix,and selects a limited number of comments to enter the next layer.Then pre-train the filtered comment set with the BERT model.And using Bi-GRU+Attention neural network to extract its deep semantics.Finally,by combining the improved probability matrix decomposition,predict the user's rating for the product,and use this method to enhance the accuracy of the recommendation system.The validity of the algorithm is confirmed by the experiment.(2)A two-way collaborative variation recommendation algorithm Bi-CVAE for comments is proposed.The algorithm attempts to model the content vector generated by the neural network into a hidden variable,and incorporates the feature-based feature vector into the classical variational automatic encoder.Then it is applied to the recommendation system through the collaborative filtering probability matrix decomposition model to improve the performance of the recommendation system.In this chapter,the VAE potential semantic generation algorithm combined with deep neural network Bi-GRU is introduced,and then the algorithm derivation of maximum posterior estimation and Bayesian reasoning is carried out in turn.Finally,the cooperative filtering algorithm based on matrix decomposition is appliedto the recommendation system to improve the accuracy of prediction.Experiments show that the algorithm is effective in improving the accuracy of prediction.(3)An online real-time recommendation system based on deep collaborative filtering is designed and implemented,and the design method of distributed data acquisition,storage and comment data analysis module is given.Using user comments as a medium and using collaborative filtering,the recommendation system can be combined with in-depth learning.The accuracy of the recommendation system is improved by updating the user feature model,and it is suitable for both offline and online recommendation task systems.