A Topic Detection Approach Through Hierarchical Clustering on Concept Graph

• DOI: 10.12785/amis/070619
• Publication Date: 2013-07-15T06:54:15Z
• Authors: Xiaohui Huang; Xiaofeng Zhang; Yunming Ye; Shengchun Deng; Xutao Li
• Keywords: Hierarchical clustering

Topic detection and tracking (TDT) algorithms have long been developed for the discovery of topics. However, most existing TDT suffer from paying less attention to: (1) temporal distance between a pair topics; (2) mutual effect highly correlated topic terms. In this paper, we proposed novel approach by applying hierarchical clustering on constructed concept graph (HCCG), which is able to solve aforementioned shortcomings simultaneously. approach, first defined as well behavior curve. Then, tempo ral with vertexes connected edges sharing same By performing graph, curves will be grouped together The evaluated number datasets promising experimental results show that our superior K-means, agglomerative algorithm(AGH), LDA respects precision, recall F-measure. Moreover, can used track change trend monitoring peak frequency curves.