社群網路分析是一種透過資訊技術來對映及測量個人、群體、組織等不同群體的方法，它能利用社交資訊或社交行為之收集，來分析出不同情境下之社群關係，近年來已經有許多人投入社群網路分析的行列。其中，以社群偵測和發現為較熱門的領域，它能找出隱藏的社群用以輔佐推薦系統。 但是我們發現很少研究提到以階層化的方式去探討對抽象化社群的影響，因此，本篇論文的目的是設計一個藉由自動化產生的文章類別階層結合使用者有興趣之文章的角色分類分法。由實驗結果顯示，我們所提出的方法確實可以藉由階層化的方式提升分類的正確性。Social network analysis is a methodology to collect, analyze, and display the community relationship under different scenarios, it utilizes varied techniques to measure the social information, user-generated content, and social interaction. The last few years have seen a great deal of work on social network analysis. Community detection and discovery particularly is the most popular filed, and it can find the hidden communities to further analysis, such as community recommendation. However, role identification is a difficult job for many social network applications. One of the difficulties is to maintain and utilize large amount of distinct roles. And we found that there are few studies of any kind have examined the influence of using concept hierarchy to social network abstraction. In this paper, we attempt to adapt fuzzy classification method and construct a hierarchy for role classification, in other words, we want to design a role classification methodology based on the documents which users are interested in, and attempts to form the role hierarchy automatically then analyzes it. We believe this approach can encourage the utilization of social roles by considering their identifiable features at different levels.
