Research And Design Of Explainable Movie Recommendation System

With the rapid development of information technology,5G has stepped into the process of commercialization,and the problem of information overload has intensified.The emergence of the recommendation system has largely alleviated this problem,which is now used as a basic technology in other industries in the field of movie,e-commerce,news and we-media.As a mainstream leisure entertainment,movies are very popular among people.As the film and television industry develops fast,a large amount of film resources have been accumulated.Although each film has its own characteristics,the preferences of users are quite different.So how to find a movie that is suitable our own taste becomes a big challenge.With the continuous development of machine learning and deep learning in the field of recommendation system,the accuracy of the system is also increasing,but only providing users with recommended results is not sufficient.Studies have shown that explaining the reasons for recommendation to users helps improve the transparency of the recommendation system and user satisfaction when providing users with recommended items.The paper focuses on the issues related to recommendation interpretation generation in the explainable movie recommendation system.This paper proposes an algorithm for generating personalized recommendation explanations based on topic words.The algorithm is mainly divided into two modules,namely,recommendation module and explanation generation.The recommendation module uses a multi-layer perceptron model.The explanation generation module applies the word frequency statistics with fused semantics.This module extracts topic words from the user comment information to generate recommended explanations combining with the gated recurrent neural networks.Experiments on the Douban movie dataset show that the algorithm performances well in terms of recommendation accuracy and recommendation explanation generation.This paper puts forward an algorithm of recommendation explanation generation that incorporates neighbor comments.The algorithm is divided into two modules,namely,recommendation algorithm and explanation algorithm.The former uses a multi-layer perceptron.The later one supplements the missing comment datas by constructing the neighbor comment dataset,and then uses the topic model to find the topic words of target users for the item from comments.Finally,the recommended explanations are generated with the pre-defined templates.This paper designs an explainable movie recommendation system based on the Spark distributed platform by using the Movie Lens-10 M public dataset.The recommendation algorithm uses a multilayer perceptron and item-based collaborative filtering.The realization of the system verifies the feasibility of the above-mentioned explainable movie recommendation algorithm.The feasibility of the above explainable movie recommendation algorithm is verified by the system implementation.