摘要: 为解决文本主题表达存在的信息缺失问题,提出一种基于词共现的文档聚类算法。利用文档集上的频繁共现词建立文档主题向量表示模型,将其应用于层次聚类算法中,并通过聚类熵寻找最优的层次划分,从而准确反映文档之间的主题相关关系。实验结果表明,该算法所获得的结果优于其他基于短语的文档层次聚类算法。
关键词:
文档聚类,
文档模型,
词共现,
文档相似度,
聚类增益
Abstract: This paper presents a document clustering algorithm based on word co-occurrence to solve the problem about information deletion of text subject expression. It uses the word co-occurrence of document set to establish the document theme vector presentation model, and applies to the hierarchical clustering algorithm, through the clustering entropy to find the best level partition, and accurately reflects the relationship between documents’ theme. Experimental results show that the algorithm results is better than other phrases document hierarchical clustering algorithm.
Key words:
document clustering,
document model,
word co-occurrence,
document similarity,
clustering gain
中图分类号:
常鹏, 冯楠, 马辉. 一种基于词共现的文档聚类算法[J]. 计算机工程, 2012, 38(2): 213-214.
CHANG Feng, FENG Nan, MA Hui. Document Clustering Algorithm Based on Word Co-occurrence[J]. Computer Engineering, 2012, 38(2): 213-214.