Design And Implementation Of Recommendation Algorithm Based On Sentiment Analysis

When using the information in the network,in order to enhance the adhesion between users and commodities,the recommendation system and personalized recommendation system have been introduced after the emergence of the search engine,and the personalized recommendation system based on sentiment analysis has gradually emerged.More and more important status.The current research focus on personalized recommendation system based on sentiment analysis focuses on the sentiment analysis module,which is how to correctly and accurately analyze the user's emotional tendency.At present,the network information is more and more comprehensive,and the types of information that can reflect the user's emotional tendency are more and more.How to effectively combine different emotion expressions,how to accurately analyze Chinese information,the Chinese text-cutting technology,the emotional quantitative model design and the combination of different emotional expression methods.In Chinese word segmentation,the LSTM model is used to deal with feature extraction and other issues.In the emotional quantification model,there are problems in the immature model,the model is more confusing.This paper has carried out research on the above aspects.The specific research contents are as follows:(1)The LSTM model has been improved to improve the accuracy of Chinese word segmentation.First,based on the model memory time problem,the model is improved to increase the memory time of the model.Secondly,for the complexity of the LSTM model,the model is improved to enhance the ability to extract features.(2)For the establishment of the emotional quantification model,the Chinese commentary is specifically emotionally quantified.Make full use of all the words of the comment text,not only the emotional words,but also the study of the negative words and degree adverbs in the sentence.The emotional value of the entire comment statement can be calculated more accurately.In order to effectively combine user comments and user ratings,and use different methods of user emotion expression,Monte Carlo method is used to calculate the different weight values of the two,and the two can be effectively combined to calculate more accurate.Affective value.(3)According to the uniqueness of the data in this paper,the improved collaborative filtering recommendation algorithm adapted to the data of this paper is selected for recommendation,so that the processed data is more fully utilized,so that the accuracy of recommendation is compared with the previous recommendation algorithm.The accuracy has been improved.In summary,this paper has done a lot of research and analysis on sentiment analysis,improved the LSTM model,designed the emotional quantitative model and conducted an effective combination of user comments and users.After simulation experiments and experimental results,the improved LSTM model has improved and improved the accuracy and recall rate of Chinese word segmentation.The designed emotional quantification model and the combination of user reviews and user ratings are used in the final recommendation algorithm.The recommendation results have been improved in accuracy and recall rate.