SELF-SUPERVISED PRODUCT FEATURE EXTRACTION USING A KNOWLEDGE BASE AND VISUAL CLUES

Rémi Ferrez, Clément de Groc, Javier Couto

2012

Abstract

This paper presents a novel approach to extract product features from large e-commerce web sites. Starting from a small set of rendered product web pages (typically 5 to 10) and a sample of their corresponding features, the proposed method automatically produces labeled examples. Those examples are then used to induce extraction rules which are finally applied to extract new product features from unseen web pages. We have carried out an evaluation on 10 major French e-commerce web sites (roughly 1 000 web pages) and have reported promising results. Moreover, experiments have shown that our method can handle web site template changes without human intervention.

Paper Citation

in Harvard Style

Ferrez R., de Groc C. and Couto J. (2012). SELF-SUPERVISED PRODUCT FEATURE EXTRACTION USING A KNOWLEDGE BASE AND VISUAL CLUES . In Proceedings of the 8th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST, ISBN 978-989-8565-08-2, pages 643-652. DOI: 10.5220/0003936706430652

in Bibtex Style

@conference{webist12,
author={Rémi Ferrez and Clément de Groc and Javier Couto},
title={SELF-SUPERVISED PRODUCT FEATURE EXTRACTION USING A KNOWLEDGE BASE AND VISUAL CLUES},
booktitle={Proceedings of the 8th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST,},
year={2012},
pages={643-652},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0003936706430652},
isbn={978-989-8565-08-2},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 8th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST,
TI - SELF-SUPERVISED PRODUCT FEATURE EXTRACTION USING A KNOWLEDGE BASE AND VISUAL CLUES
SN - 978-989-8565-08-2
AU - Ferrez R.
AU - de Groc C.
AU - Couto J.
PY - 2012
SP - 643
EP - 652
DO - 10.5220/0003936706430652