User Tools

Site Tools


A comparative study on the use of similarity measures in case-based reasoning to improve the classification of environmental system situations

Héctor Núñez, Miquel Sàchez-Marrèà, Ulises Cortés, Joaquim Comas, Montse Martíez, Ignasi Rodríguez-Roda, Manel Poch

For the management of environmental system (ES) is used the environmental decision support system (EDSS). It is curtail for EDSS to make reliable diagnostics and proposed plan to deal with the situation. The diagnosis starts with determination of the key parameters (i.e., group of variables and their interrelations) that identify an operational state. Physical, biological and chemical indicators (numerical and/or symbolic) transform data and statistics into synthetic information. The initial diagnosis determines if the process is in a normal or abnormal state. The objective of the paper is to apply CBR for deeper classification task in order to make diagnosis based on previous experience and generate a sophisticated action plan. In addition, some environmental problems can only be solved by experts using their own experience in the resolution of similar situations. These experts are not always accessible when dealing with risk situations, and it is crucial to record each new experience to learn about the process, while reusing this specific knowledge in the future. The article discusses similarity measures (Minkowski’s metrics, unweighted and heterogeneous similarity measures) and introduces L’Eixample - normalized weight-sensitive similarity measure, which “[…] takes into account the different nature of the quantitative or qualitative values of the continuous attributes depending on its relevance”. There is a test performed, which test the efficiency of similarity measures.

kritis/point_of_view.txt · Last modified: 2015/03/24 15:32 (external edit)