Recommendation Algorithm Based On The Content And Emotion Of Weibo

With the rapid development of the Internet,Weibo is becoming more and more popular.The information of users has an exponential growth,and the massive content has caused people to face serious information overload.While the recommendation system can mine user preferences,filter invalid information,and improve search efficiency.So,it can be used as an effective means to solve the problem of information overload,greatly improving the user experience.In addition,The interactivity and immediacy of the Weibo content also provides the basis for the recommendation system.However,the traditional recommendation algorithm is not well suited for Weibo because it may ignore the emotions of the user's published content.Therefore,adding emotional factors to the recommendation algorithm has become a new research direction.In this paper,the sentiment analysis algorithm and recommendation algorithm are studied with Weibo content as the research object.The main research contents are as follows:(1)For the case that the sentiment dictionary cannot adapt to the complex semantics of the Weibo content,the network new words are collected and added to emotion dictionary.And,in view of the fact that the emoticons usage rate of expressions for Weibo content is getting higher and higher,the emoticons are added to emotion dictionary.What's more,the traditional SVM sentiment analysis algorithm can't solve the problem of short text noise.So,the structured emotion dictionary is used to extract feature items to remove the noise without emotion information.Then,according to the TF-IDF vectorization feature as the input,the SVM model is trained to discriminate the emotional polarity of the Weibo content.And the optimal performance of the classifier is obtained by experimenting with changing the parameters of the SVM.Finally,in order to complete the multi-level emotional scoring,the adverb of degree is extracted to calculate the emotional intensity value,which is used in the recommendation algorithm to obtain the final emotional score.(2)For the situation that Weibo's recommendation method is only based onpopularity recommendations,a personalized recommendation scheme is proposed.It is using the user-based collaborative filtering algorithm to give a recommendation list.Weibo does not have a scoring system that cannot directly use the collaborative filtering algorithm,and can only use the implicit recommendation algorithm.This paper uses Key words extraction technology to extract the keywords of the Weibo content as the item of the project,and relies on the result of sentiment analysis as the scores to form the item-scoring matrix.Then,the final similarity is got by weighted fusion calculating the cosine similarity and the Gerrard coefficient.And the recommendation list is given according to the prediction score.And by changing the parameters of the algorithm,the accuracy of the recommendation algorithm is improved.At the same time,the basic information of the user is used to solve the cold start problem according to the popularity recommendation.The experimental results show that the emotion discriminant algorithm based on sentiment dictionary optimization SVM can effectively improve the classification accuracy of emotional polarity.And it can solve the problem of short text,high noise and complex semantics of Weibo content.At the same time,the user-based improved collaborative filtering algorithm can effectively solve the problem that the Weibo content has no scoring system.The similarity fusion calculation helps to improve the performance of the recommendation system and to obtain better recommendation results.