Research And Implementation Of Book Recommendation System Integrating Reviews And Ratings

With the increase of people’s demand for books,book recommendation systems expand the limitations of users’ choice of books and provide a wider range of reading choices.However,due to the huge number of books,it is difficult for users to quickly find the books they are interested in.Researchers have applied personalized recommendation technology to the recommendation system,which highly improves the efficiency of user selection and purchase.However,the traditional book recommendation system is mainly based on the user’s historical transaction behavior data,ignoring the user’s feedback information such as reviews of books.At the same time,the existing book recommendation systems also have problems such as inaccurate recommendation results caused by sparse user rating matrix.Considering that reviews have emotional information,learning the representations of users and books through reviews can not only alleviate the sparsity problem of the rating matrix,but also improve the interpretability of the recommendation to a certain extent.In view of the above requirements,this paper proposes a recommendation algorithm model based on fusion reviews and fusion ratings(FRIFR),and applies the model to the developed book recommendation system,mainly completes the following work:Firstly,FRIFR model is proposed based on Deep Co NN.The algorithm considers the sentiment polarity of highly emotionally differentiated reviews(strongly positive or negative reviews)to help determine useful reviews,so the emotional attention mechanism and global attention mechanism are introduced to help identify user reviews that are more likely to improve the accuracy of rating prediction model.Considering that the rating matrix has the potential correlation between users and books,the DMF model is used to obtain the characteristics of users and books.Then,the features obtained from reviews and ratings are respectively used by LFM algorithm to obtain the predicted value.Finally,the predicted values were dynamically fused to output the final predicted value,and the recommended list was generated according to the final predicted value to recommend books to users.The experimental results show that the proposed algorithm model is more accurate than other algorithms in RMSE and MAE,and can achieve more accurate recommendation effect.According to the requirements of the development of book recommendation system based on FRIFR algorithm,through the investigation of the existing book recommendation system,first from the book recommendation system user needs and system functional requirements analysis to design the overall framework of the system,and then further design the detailed functional modules and database,and finally complete the coding and testing of the whole system.The book recommendation system can meet the basic functions of users such as purchase,browse,review and so on.At the same time,the FRIFR recommendation algorithm proposed in this paper can recommend books that are more interesting to users.