Issues concerning social network analysis have drawn significant attentions from the public. One of the most eye-catching issues is the gradually-improving studies regarding text analysis on social network services. Studies pointed out that users' emotion status can be recognized, and modeled as well, through sentiment analysis, and furthermore, prompted social awareness such as trend/opportunity exploration, business strategy development, and so on. It is found that not only the text, e.g., tweet, post, etc., itself may lead to the changes in users' emotion but also any bi-directional activities during the whole social interaction process. Considering emotion as the result of human judgment on the overall feeling in the cognitive theory of emotion, this study targets to design a universal model that supports the monitoring of users' emotion change on social network service. This model employs a new multi-dimension mapping method to detect and analyze users' emotion changes based on machine learning technique, and target to prompt further understanding of human behavior in online society.