###
Journal of Software:2012.23(6):1486-1499

一种结合主动学习的半监督文档聚类算法
赵卫中,马慧芳,李志清,史忠植
(湘潭大学 信息工程学院,湖南 湘潭 411105;中国科学院 计算技术研究所 智能信息处理重点实验室,北京 100190;西北师范大学 数学与信息科学学院,甘肃 兰州 730070;中国科学院 计算技术研究所 智能信息处理重点实验室,北京 100190)
Efficiently Active Learning for Semi-Supervised Document Clustering
ZHAO Wei-Zhong,MA Hui-Fang,LI Zhi-Qing,SHI Zhong-Zhi
(College of Information Engineering, Xiangtan University, Xiangtan 411105, China;Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, The Chinese Academy of Sciences, Beijing 100190, China;College of Mathematics and Information, Northwest Normal University, Lanzhou 730070, China;Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, The Chinese Academy of Sciences, Beijing 100190, China)
Abstract
Chart / table
Reference
Similar Articles
Article :Browse 3889   Download 5020
Received:November 03, 2010    Revised:March 31, 2011
> 中文摘要: 半监督文档聚类,即利用少量具有监督信息的数据来辅助无监督文档聚类,近几年来逐渐成为机器学习和数据挖掘领域研究的热点问题.由于获取大量监督信息费时费力,因此,国内外学者考虑如何获得少量但对聚类性能提高显著的监督信息.提出一种结合主动学习的半监督文档聚类算法,通过引入成对约束信息指导DBSCAN的聚类过程来提高聚类性能,得到一种半监督文档聚类算法Cons-DBSCAN.通过对约束集中所含信息量的衡量和对DBSCAN 算法本身的分析,提出了一种启发式的主动学习算法,能够选取含信息量大的成对约束集,从而能够更高效地辅助半监督文档聚类.实验结果表明,所提出的算法能够高效地进行文档聚类.通过主动学习算法获得的成对约束集,能够显著地提高聚类性能.并且,算法的性能优于两个代表性的结合主动学习的半监督聚类算法.
Abstract:Semi-Supervised document clustering and employing limited prior knowledge to aid in unsupervised clustering, have recently become a topic of significant interest to data mining and machine learning communities. Because receiving supervised data may be expensive, it is important to attain the most informative knowledge to improve the clustering performance. This paper presents a semi-supervised document clustering algorithm with active learning for pairwise constraints, aiming at getting improved clustering performance. The semi-supervised document clustering algorithm is a constrained DBSCAN (cons-DBSCAN) algorithm, which incorporates pairwise constraints to guide the clustering process in DBSCAN. Basing on measure of constraint set utility and analysis of DBSCAN algorithm, an active learning approach is proposed to select informative document pairs for obtaining user feedbacks. Experimental results show that this proposed approach is effective in document clustering. The clustering performance of active Cons-DBSCAN has dramatically improved with selected pairwise constraints. Moreover, the proposed approach performs better than the two representative methods.
文章编号:     中图分类号:    文献标志码:
基金项目:国家自然科学基金(61105052, 61070232); 湖南省自然科学基金(11JJ4051); 湖南省教育厅一般项目(10C1262); 湘潭大学博士启动基金(10QDZ42); 中国科学院计算技术研究所智能信息处理重点实验室开放基金(IIP2010-6); 西北师范大学青年教师科研能力提升计划骨干项目(NWNU- LKQN-10-1) 国家自然科学基金(61105052, 61070232); 湖南省自然科学基金(11JJ4051); 湖南省教育厅一般项目(10C1262); 湘潭大学博士启动基金(10QDZ42); 中国科学院计算技术研究所智能信息处理重点实验室开放基金(IIP2010-6); 西北师范大学青年教师科研能力提升计划骨干项目(NWNU- LKQN-10-1)
Foundation items:
Reference text:

赵卫中,马慧芳,李志清,史忠植.一种结合主动学习的半监督文档聚类算法.软件学报,2012,23(6):1486-1499

ZHAO Wei-Zhong,MA Hui-Fang,LI Zhi-Qing,SHI Zhong-Zhi.Efficiently Active Learning for Semi-Supervised Document Clustering.Journal of Software,2012,23(6):1486-1499