Topic-based Conformity Influence Modeling And Analysis In Social Network

With the rapid development of information technology, the links between social networks and people’s lives are becoming more and more close, and social networks become an irreplaceable part in people’s lives. Where in the social network, the impact study among people is a very important issue, and people being affected by the impact from the surrounding user’s own characteristics in the network. How to reflect these features to the social network model is an important research direction. Because the more accurate quantification, and its implications for people’s lives is larger, such as marketing, voter voting predicted, to maximize the social impact of the application, etc., in short, precise quantification social network model is a very interesting research.When building the social network model, especially in the construction of the user when the herd of the algorithm, most of the existing algorithms do not consider the influence of different users that under the theme is different, that people affected by different environments will change their choices. Based on LDA topic model, a herd of the model based on the effects of different topics, the model calculates the herd effects model under different themes, and then take their weight as a comprehensive herd model. Finally, this general model is used to describe the social networks impact models. We apply the model to predict the behavior of the user, and then it is compared with some previous algorithms, the experimental results show that our algorithm to predict the effect due to others, indicating conformity of the model in this paper that is effective and meaningful.As our model takes into account the topic factors, so to the application of our model, we can solve some practical problems. This article lists two problems. The first is to find a topic most likely to select a user behavior. We consider part of our herd utility function model is a weighted sum of the individual utility function under the theme, so we can use these fractional considered separately by calculating the same subject with a different user behavior utility function, we utility the value of the function to find the user, the user that is which we looking for. The second type of problem is to find a user selects a theme that is most affected behavior. We will still divide the utility function into various fractions utility function by calculating a user with the same behavior utility function under different themes, we find that subject by the size of the value of the utility function, the subject is the theme we’re looking for.