Feature Extraction Using Restricted Bootstrapping - Collections

＜conference paper＞
Feature Extraction Using Restricted Bootstrapping

Creator	Creator Name 廣川, 佐千男 Hirokawa, Sachio Affiliation Affiliation Name 九州大学情報基板研究開発センター Research Institute for Information Technology, Kyushu University
Language	English
Publisher	IEEE
Date	2012-05
Source Title	2012 IEEE/ACIS 11th International Conference on Computer and Information Science (ICIS)
First Page	283
Last Page	288
Publication Type	Accepted Manuscript
Access Rights	open access
Related DOI	http://doi.org/10.1109/ICIS.2012.50
Related URI	http://doi.org/10.1109/ICIS.2012.50
Relation	http://doi.org/10.1109/ICIS.2012.50
Abstract	The bootstrapping method is known as an application of the ranking technique of Google for documents and words. The technique calculates a importance of the word by propagating the score of the word a...nd the document mutually. However, sometimes the result is far away from the query word. The problem is known as "Topic drift". This paper proposes the restriction to high-ranking words in the process of bootstrapping. The method is simpler than the technique so far and generalized method such as limiting the emerging pattern of the word. It is confirmed that the method was able to prevent the topic drift, and to extract the bankruptcy reason strongly related to the query word the real bankruptcy information documents.show more

File	FileType	Size	Views	Description
hirokawa_83	pdf	74.3 KB	350

Record ID	1546804
Peer-Reviewed	Refereed
Subject Terms	topic drift
	feature word
	bankruptcy information
	text mining
ISBN	978-1-4673-1536-4
DOI	10.1109/ICIS.2012.50
Notes	2012 IEEE/ACIS 11th International Conference on Computer and Information Science (ICIS) : 30 May - 01 Jun 2012 : Shanghai, China
Created Date	2015.12.10
Modified Date	2016.03.23