Identification of significant business cycle indicators using query logs of search engine
Date Issued
2010
Date
2010
Author(s)
Chen, Jeng-Chien
Abstract
Identifying status of business cycles is critical to governments and enterprises when building business strategies. Traditionally, economic variables, such as industrial production, stock price index, manufacturing sales, are selected to compose business cycle indicators, which altogether evaluate business cycle status. In general, the release of economic variables involves long data processes that delay the announcement of business status. The announced business status thus is not timely and could increase the uncertainty of business decision making. In this work, we employ query logs of search engines for business cycle identification. As query logs are readily available through online Web services, they can provide timely and accurate information about business status. We propose a feature selection method to identify query terms appropriate for business cycle identification. Evaluation results show that the identified query terms are effective indicators and the proposed method models business cycles correctly.
Subjects
text mining
business indicator
business intelligence
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-99-R97725017-1.pdf
Size
23.32 KB
Format
Adobe PDF
Checksum
(MD5):4422adfff3c315192fa093601336d338