The Design And Implementation Of A Personalized News Recommendation System Based On Twitter

Information recommendation system can extract useful information from massive data and push it to users,which is used to solve the problem of "information overload" in the era of big data.In order to solve the problem that users can't get news according to the news reading habits and the needs of the current scene,a personalized news recommendation system based on Twitter is proposed.This system recommends personalized news to the user through the user's reading habits.This system uses the B/S mode.It's main structure consists of the data display layer,the data processing layer,and the data storage layer.It's main functions consists of news collection,user information gathering,data mining,news recommendation and user interest feature management.The news collection function is in charge of crawling news from Twitter and preliminary processing of the data.Information recommendation function includes hot spot recommendation,personalized recommendation and focused user recommendation.The news collection function is in charge of crawling data from Twitter and calculating the topic distribution of the news,calculating the heat and storing data.The user information collection function is responsible for obtaining the user's behavior records and personal data.Data mining is responsible for news topic modeling,user feature computing,and similar user computing.The news recommendation function establishes user characteristics based on users' reading,likes,comments on the news to provide users with personalized news information.User interest feature management function is responsible for adding or deleting topics and lexical labels to modify user features.The database of personalized news recommendation system is designed.This system needs build the BTM theme classification model first for the Twitter information,then use the BTM model to calculate the theme distribution and the heat.User features and user similarity are required before the news recommendation.This system use way of mixed content and collaborative filtering recommendation.User interest features are used for news content recommendation,similar users which have similar interest features are used for collaborative filtering recommendation,and then users select the recommended scheme according to the proportion to generate the recommended results.The results of function test,performance test and usability test show that this system can provide good personalized news recommendation service.