近年來，網路的快速發展與社群網路大量個崛起，造成現代人越來越依賴網路社群等相關軟體。過去在相關的社群探勘方面的研究，不外乎以K-means演算法的變形、凝聚法、圖形化的方式抑或是建立在Girvan-Newman所提出的演算法架構之下。其中凝聚法往往搭配著核心節點與區域社群的概念使用，其中的癥結點在於，利用核心節點與區域社群概念的凝聚法，往往會忽略掉在社群邊緣的節點，進而在做最後的分配時，未能將其作妥善的分群。因此，本研究基於現有的凝聚法相關研究，找出居中度及相關性作為新的凝聚依據，根據此二指標將社群作出妥善的分群，並與(Lim & Datta, 2013; Qiong & Ting-Ting, 2010; Tiantian & Bin, 2012)等學者我提出的方法作比較，進而證明其改善之效果。;Quick development of the Internet and huge explosion of the social network make people rely highly on the social network software in their daily life. Most researches on community detection in the past refer to K-means, agglomerative, graph or Girvan- Newman algorithm. The interest of this study has been directed to the algorithm of agglomerative. One possible deficiency of this method is that it always ignores the nodes which are on the edge of the community. Therefore, in the merging step, the nodes on the edge could be allocated to the wrong community. This study is aimed to improve the performance of the algorithm by finding the core node and the local community as new indexes for agglomerate. In the experiments, the results are compared with (Lim & Datta, 2013; Qiong & Ting-Ting, 2010; Tiantian & Bin, 2012) to show the effectiveness of the method developed in this study.
