The Research Of Microblog Users' Retweet Prediction Based On Retweet Habits

With the increasing scale of microblog users,microblog has gradually become an important carrier of information dissemination.The main way of microblog information dissemination is retweet.Predicting microblog retweet can help understand the way of microblog information dissemination,and it is of great significance to the application of network marketing and public opinion monitoring.Most of the existing microblog users' retweet prediction methods consider the influencing factors of forwarding behavior from two perspectives: the forwarder and the publisher.The effect of influencing factors considered from the perspective of the forwarder is better,but the accurate extraction of influencing factors still needs to be improved.The influencing factors considered from the perspective of publishers do not reflect the individual differences between forwarders,and the effect of retweet prediction for specific users is not ideal.In order to solve this problem,a prediction method based on retweet habits is proposed.First of all,improving the influencing factors and mining users' retweet habits from three aspects of retweet behavior habits,retweet object habits and retweet content habits,more accurately describing users' retweeting characteristics.Then,the classification algorithm of machine learning is used to predict the microblog retweet,and the validity of the method is verified by experiments.The main work of this paper includes two aspects:(1)Mining users' retweet habits.First of all,it analyzes the influencing factors of retweet,and divides them into three aspects: users' behavior characteristics,users' social inclination and users' content preference.Then,the concept of retweet habits is proposed,and the mining methods of retweet behavior habits,retweet object habits and retweet content habits are given respectively.Among them,the retweet behavior habit considers two aspects:the user's own retweet behavior preference and the retweet activity in the whole microblog network;the retweet object habit considers three aspects: the social relationship between the user and other users,the number of fans preference and the authentication attribute preference;the retweet content habit considers two aspects: the user's content interest and content attribute preference.(2)Achieve the prediction of retweet microblog.First,the corresponding retweet habits are quantified by each information to be predicted(including the information of microblog to be predicted and its publisher),and the corresponding prediction components are obtained.Then,the prediction component is used to train the classification algorithm of machine learning,so as to realize the prediction of retweet microblog.Finally,the validity of the real data set crawled from Sina Weibo is verified.Two groups of comparative experiments are set up:(1)the contribution of different retweet habits to microblog retweet prediction is verified;(2)the microblog behavior prediction method proposed in this paper is compared with the most classic method at present.The experimental results show that among the three microblog habits,the microblog content habit has the highest contribution to the prediction accuracy,and the method considering the three microblog habits has the best effect,which is about 2.7% higher than the classical method.