A Study Of Hybrid Personalized Recommendation Algorithm Based On Text Sentiment Analysis

With the rapid development of the Internet,web2.0 applications have sprung up,people can swim in the virtual world built in the Internet.With the greatness of material life,people began to gradually pursue the spiritual realm of ascension.A variety of virtual communities can help people to release the real life of the pressure,and help people get information.People can use the virtual community applications to express their views.A long time,the Internet has accumulated a lot of information.The advent of the tourism virtual community has helped a large number of users to inquire about travel information.At the same time,the tourism virtual community can also use the comments and information left by the people to analyze the tourist's interest.In this paper,the tourism virtual community as a research background,use the existed algorithms to analyze the user's travel comment information.The development of the Internet makes us in the "information explosion" era,people can not query from a large number of information to get the required content,quickly and accurately.The search engine has solved the problem of information overload.The search engine needs users to provide the search keywords.However,when the user can not provide accurate description of their own needs keywords,the search engine can do nothing.In order to meet the user's personalized service needs,the researchers put forward the recommendation system.At present,the research of the recommended system has gradually become more mature.But,there are still some limitations in some applications.The recommended effect of the traditional collaborative filtering algorithm is not very good,because the model does not join the user's interest information and the user's social information.In this paper,the tourism virtual community as the research background,the recommended algorithm was applied to the actual needs.On the basis of the traditional collaborative filtering model,we add the user's interest information and the user's social information.And,use the weighted method to form a new personalized recommendation model to improve the accuracy of the recommended model,improve user loyalty,ultimately.The main contents and research results of this thesis are as follows:1.The article reviews the virtual community,emotion analysis and recommendation algorithm,and reviews these literatures.Most of the research objects in the virtual community focus on the members of the community.Most of the research conducted a questionnaire survey on the members to analyze the questions of the community existence mechanism and consumer satisfaction.There are few studies on the discovery and measurement of the value of unstructured information generated by members in the community,and it did not a combine the theoretical knowledge and practical business needs to put forward some practical solutions.For the tourism recommendation model,many articles use a simple collaborative filtering model.There is no comparison between the models,and the amount of experimental data is relatively small.So that we can not know the accuracy of its recommended results.2.In order to study the user's comment information in the tourism virtual community,this paper introduces the text mining technology.The value of unstructured information generated by users in the community is studied by emotion recognition of text review information.The emotion classification algorithm based on machine learning is used to identify the emotional tendencies of the user's Chinese text review information,to obtain the interest value between the user and the tourism project.With the help of the user's interest information for personalized recommendation to improve the accuracy of traditional collaborative filtering recommendation algorithm.3.In the real world,people will be more trust in their familiar friends to give recommendations.This article introduces the user's social information into the recommendation algorithm,in order to improve the user's satisfaction,to achieve high-quality personalized recommendation.In this paper,the use's interest information is used to calculate the similarity of the user's interest,and the user's social information is used to calculate the familiarity between the users.After assigning weight to user similarity and user familiarity,combined into weighted mixed recommendation model.The paper also analyzes the evaluation of the off-line experiments of the various recommended algorithms in the actual data set.Compared with other recommended algorithms,the experimental results verify that the weighted hybrid recommendation algorithm has higher recommendation accuracy and expansibility.The innovation of this paper has the following points:(1)Use of the emotion analysis method based on machine learning to obtain the user's interest in the tourism project.Then add the use's interest information to the traditional collaborative filtering recommendation algorithm,to achieve personalized recommendation,to meet the actual needs of users.(2)In this paper,the user's interest information is used to calculate the similarity of the user's interest,and the user's social information is used to calculate the familiarity between the users.After assigning weight to user similarity and user familiarity,combined into weighted mixed recommendation model.The experimental results also prove the superiority of the proposed model.