A Probabilistic Model for Information Diffusion in Social Networks: Insights From Twitter Data

Social networks have become a part of the daily lives of most people and are a significant influence in a variety of fields including economics, culture, and politics. This has motivated research on social networks. One aspect is evaluating the impact of messages on society which is a function of the information spread. Thus, a probabilistic model is proposed for the information spread in a social network. In this model, the probability of a user retweeting is based on metrics such as the trending degree, the importance of users retweeting the message, message freshness, and the influence of users on each other. The message viewing time is also considered as it is a critical spread factor. We propose three algorithms based on the proposed model. The first is based on a Kalman filter that simply estimates the number of retweeting users in the near future. The second considers in what order and at what times users retweet the message. The third employs some simplifications to reduce the complexity of the second algorithm. A real Twitter dataset is used to evaluate the performance. The results obtained show that the proposed model accurately predicts the number of users who spread a message.