摘要: 在分析灾害新闻特点的基础上,提出一种基于文本挖掘的话题发现技术,采用基于平均分组的层次聚类算法,对灾害新闻资料进行组织,从而生成新闻专题,为用户提供个性化服务,并形成专题检测系统,同时介绍基于时间和地点权值向量的相似度计算模型以及基于时间的动态阈值模型。实验结果表明,该算法能够获得较好的性能。
关键词:
话题发现与跟踪,
层次聚类,
文本挖掘,
动态阈值
Abstract: On basis of analyzing the character of disasters news, a topic detection technique based on text mining is proposed, which uses Group Average Clustering(GAC) algorithm to organize the disasters news materials, generate the news special topics, provide the personality service, and shape the whole system. The similarity computing model based on both weight vectors of time and place and dynamic threshold model based on time are introduced. Experimental results show this algorithm can obtain better performance.
Key words:
topic detection and tracking,
hierarchical clustering,
text mining,
dynamic threshold
中图分类号:
高 妮;周明全;耿国华;王学松;贺毅岳. 基于文本挖掘的话题发现技术[J]. 计算机工程, 2009, 35(19): 36-38.
GAO Ni; ZHOU Ming-quan; GENG Guo-hua; WANG Xue-song; HE Yi-yue. Topic Detection Technique Based on Text Mining[J]. Computer Engineering, 2009, 35(19): 36-38.