|
Publication
|
Title of Article |
Information extraction for deep web using repetitive subject pattern |
Date of Acceptance |
23 July 2013 |
Journal |
Title of Journal |
World Wide Web |
Standard |
|
Institute of Journal |
Springer US |
ISBN/ISSN |
1573-1413 |
Volume |
|
Issue |
|
Month |
|
Year of Publication |
2013 |
Page |
1-31 |
Abstract |
In this paper, we propose an information extraction (IE) system for extracting data records from semi-structured documents on the Deep Web using a promising proposed technique, called Repetitive Subject Pattern. This technique was based on the hypothesis that data records in the web page must have a subject item, and the repetitive pattern of the subject items can be used to identify the boundary of data records. The system consists of four automatic tasks: (1) parsing a sample page to a DOM tree, (2) recognizing a subject string in the DOM tree, (3) using the subject string for identifying the pattern of data records and generating a wrapper, and (4) using the generated wrapper for extracting data records. This approach enables the very flexible wrapper generator; when the automatic process generated the wrong wrapper, user can also provide a new sample subject string for generating better wrapper. As the result, the system can be both semi-supervised and unsupervised system. The experimentation shows that the proposed technique provides the outstanding results in generating the very high quality wrappers, with both recall and precision close to 100 % when tested on a number of datasets. |
Keyword |
Information extraction,Web data extraction, Web content mining, Subject pattern, Wrapper induction, Unsupervised learning |
Author |
|
Reviewing Status |
มีผู้ประเมินอิสระ |
Status |
ตีพิมพ์แล้ว |
Level of Publication |
นานาชาติ |
citation |
false |
Part of thesis |
true |
Attach file |
|
Citation |
1
|
|
|
|
|
|
|