Using entropy-based local weighting to improve similarity assessment

This paper enhances and analyses the power of local weighted similarity measures. The paper proposes a new entropy-based local weighting algorithm to be used in similarity assessment to improve the performance of the CBR retrieval task. It has been carried out a comparative analysis of the performan...

Descripción completa

Detalles Bibliográficos
Autores: Núñez, Héctor, Sànchez-Marrè, Miquel|||0000-0001-9848-5779, Cortés García, Claudio Ulises|||0000-0003-0192-3096, Comas, Joaquim, Rodriguez-Roda, Ignasi, Poch, Manel
Tipo de recurso: informe técnico
Fecha de publicación:2002
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/97483
Acceso en línea:https://hdl.handle.net/2117/97483
Access Level:acceso abierto
Palabra clave:Entropy
Similarity
CBR
UCI machine learning database repository
Local weighted similarity measures
Àrees temàtiques de la UPC::Informàtica
Descripción
Sumario:This paper enhances and analyses the power of local weighted similarity measures. The paper proposes a new entropy-based local weighting algorithm to be used in similarity assessment to improve the performance of the CBR retrieval task. It has been carried out a comparative analysis of the performance of unweighted similarity measures, global weighted similarity measures, and local weighting similarity measures. The testing has been done using several similarity measures, and some data sets from the UCI Machine Learning Database Repository and other environmental databases.