| Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 102 === Since the tremendous success of social networking websites, the research on related analysis have widely been studied. Among these research, the social influence is a significant and popular topic. We rely on the social influence model to predict and learn the influence diffusion process. There are several influence models proposed before. However, traditional models only categorize nodes into two types of state, active and inactive. This strategy causes some problems. Previous models only take the positive influence into account. In addition, if the inactive nodes are influenced successfully as active nodes, these active nodes could not return to inactive nodes. In this work, we not only improve above limitations but also propose an novel propagation method in our model. Our model proposes five states to represent the multiple states of influence. By our propagation method, the strength of social influence might be reduced when time elapsed. Eventually, we utilize the measurement of precisions to compare with related models. Our model outperforms than other models with two states. In addition. we predict the result of five states using our model. By means of experiment result, we indicate the importance of multiple states.
|
|---|
