DMCA
Web Directories as Training Data for Automated Metadata Extraction Martin Kavale2 Vojte h Svate and Pe r Strossa (2001)
| Venue: | Semantic Web Mining, Workshop at ECML/PKDD--2001, Freiburg 2001 |
BibTeX
@INPROCEEDINGS{Ormation01webdirectories,
author = {Departmentof Inf Ormation and Martin Kavalec and Vojtěch Svátek and Petr Strossa},
title = {Web Directories as Training Data for Automated Metadata Extraction Martin Kavale2 Vojte h Svate and Pe r Strossa},
booktitle = {Semantic Web Mining, Workshop at ECML/PKDD--2001, Freiburg 2001},
year = {2001},
pages = {39--44}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Introduction Although man-made annotations are conside e asthe main `knowleL e fue2 for the Se anticWe , the majority of eW sting comme rcial page are still poorly e[H-7 e with any kind of meL data,ne e r mind the forthcoming standards such as the RDF syntax or the DublinCore se antics. Infor ion Extr action, lying on characte istic patte ns in teL# can be applie e ve n on such`le acy' page- in orde r to obtain me adata containing, fore xample , the name s, type s, and domains of activity of the WWW subjefi s (companie s). The twodeH2## of deeL[fi;H t of Information Extraction tehnique have shown thate xtraction patte rns applicable onrefi-world, unstructure te[ data cannot be satisfactorily preilyW by hand;instefi; Machine Le[-2fi[ (ML) b e came the e nabling te hnology. The common assumption in ML-base Information Extraction is that the training case are chunks of tefi pre; ab e le d by a human inde;H] This isacceL-WTfi for domain-spemai reain-sp withlimite vocabulary andmoreH2WTLHH
Keyphrases
pe strossa training data web directory machine le applicable onrefi-world man-made annotation human inde unstructure te data cannot thate xtraction patte forthcoming standard dublincore se antic mel data se anticwe infor ion extr action deel fi information extraction tehnique applie ve ml-base information extraction tefi pre te hnology adata containing domain-spemai reain-sp withlimite vocabulary andmoreh2wtlhh comme rcial page knowlel fue2 le acy training case rdf syntax www subjefi common assumption characte istic patte n
