Attribute Value Ontology - Using Semantics in Data Mining

Tomasz Łukaszewski, Joanna Józefowska, Agnieszka Ławrynowicz

2012

Abstract

We propose a new concept to represent attribute values as an ontology that allows modeling different levels of abstraction. In this way more or less precise values may be used instead of missing or erroneous data. The goal is to use this representation in order to improve analysis of imperfect data. The proposed attribute value ontology (AVO) allows to upgrade the precision of information not only from positive observations but also from negative ones. We show how to classify a new example using a set of training examples described in the same or more precise way. Another advantage of the proposed approach is providing an efficient way to avoid the effect of overfitting.

Paper Citation

in Harvard Style

Łukaszewski T., Józefowska J. and Ławrynowicz A. (2012). Attribute Value Ontology - Using Semantics in Data Mining . In Proceedings of the 14th International Conference on Enterprise Information Systems - Volume 2: SCOE, (ICEIS 2012) ISBN 978-989-8565-11-2, pages 329-334. DOI: 10.5220/0004157003290334

in Bibtex Style

@conference{scoe12,
author={Tomasz Łukaszewski and Joanna Józefowska and Agnieszka Ławrynowicz},
title={Attribute Value Ontology - Using Semantics in Data Mining},
booktitle={Proceedings of the 14th International Conference on Enterprise Information Systems - Volume 2: SCOE, (ICEIS 2012)},
year={2012},
pages={329-334},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0004157003290334},
isbn={978-989-8565-11-2},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 14th International Conference on Enterprise Information Systems - Volume 2: SCOE, (ICEIS 2012)
TI - Attribute Value Ontology - Using Semantics in Data Mining
SN - 978-989-8565-11-2
AU - Łukaszewski T.
AU - Józefowska J.
AU - Ławrynowicz A.
PY - 2012
SP - 329
EP - 334
DO - 10.5220/0004157003290334