Topic discovery of clusters from documents with geographical location

doi:10.1002/cpe.3474

CORC > 计算技术研究所 > 中国科学院计算技术研究所 > 中国科学院计算技术研究所期刊论文 > 英文

	Topic discovery of clusters from documents with geographical location
	Zhang, Li 2,3; Sun, Xiaoping 2,4; Hai Zhuge 1,2
刊名	CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
	2015-10-01
卷号	27 期号:15 页码:4015-4038
关键词	topic model clustering textual document geographical location
ISSN号	1532-0626
DOI	10.1002/cpe.3474
英文摘要	As smart phones with GPS become popular, more and more textual documents with geographical locations are published on the Web. Keyword-based location services like vehicle navigation, tour planning, nearby object querying, and location pattern discovering of spatial objects are becoming popular and important. However, processing both text and geographical locations brings more challenges to information-retrieval techniques. In this paper, we focus on the problem of finding textual topics of clusters containing spatial objects with text descriptions. The key is how to combine clustering techniques with topic-retrieval models to integrate both geo-location information and text information. We investigated methods that combine clustering methods with the Latent Dirichlet Allocation model to discover topics of clusters of documents with geo-locations. Six different methods of combination are investigated, each having outputs with different meanings, which can be further leveraged to answer different types of queries over spatial documents. Experiments are conducted on both synthetic and real data. The results show that the combination of the probabilistic topic model with clustering algorithms is an efficient and effective way to discover meaningful clusters in different facets and levels of documents with textual and geographical information. Copyright (C) 2015 John Wiley & Sons, Ltd.
资助项目	National Science Foundation of China[61075074] ; National Science Foundation of China[61070183] ; Open Fund Project of State Key Laboratory of Software Engineering (Wuhan University, China)[SKLSE2012-09-2]
WOS研究方向	Computer Science
语种	英语
出版者	WILEY-BLACKWELL
WOS记录号	WOS:000363042100014
内容类型	期刊论文
源URL	[http://119.78.100.204/handle/2XEOYT63/9210]
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Sun, Xiaoping
作者单位	1.Nanjing Univ Posts & Telecommun, Nanjing, Jiangsu, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, Knowledge Grid Lab, Key Lab Intelligent Informat Proc, Beijing, Peoples R China 3.Univ Chinese Acad Sci, Beijing, Peoples R China 4.Wuhan Univ, State Key Lab Software Engn, Wuhan 430072, Hubei, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Li,Sun, Xiaoping,Hai Zhuge. Topic discovery of clusters from documents with geographical location[J]. CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE,2015,27(15):4015-4038.
APA	Zhang, Li,Sun, Xiaoping,&Hai Zhuge.(2015).Topic discovery of clusters from documents with geographical location.CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE,27(15),4015-4038.
MLA	Zhang, Li,et al."Topic discovery of clusters from documents with geographical location".CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE 27.15(2015):4015-4038.