AAAI Publications, Twenty-Third International FLAIRS Conference

Font Size: 
Similarity Measures in Hierarchical Behaviours from a Structural Point of View
Gonzalo Florez-Puga, Belen Diaz-Agudo, Pedro Gonzalez-Calero

Last modified: 2010-05-06

Abstract


Case-Based Reasoning (CBR) systems dealing with complex object-based case representation structures need to employ complex structured-based similarity measures. However, obtaining such similarity requires to solve problems on graphs are known to be NP-complete. In this paper, we show that, in spite of its theoretical complexity, structured-based similarity is of practical use and can be incorporated into the CBR toolbox. We analyze, in terms of quality and efficiency, three different methods for assessing similarity between graphs, which we apply in the domain of behaviour generation for a soccer simulation environment (SoccerBots). Our implementation of such methods has been incorporated into jCOLIBRI, a general framework for CBR development, and ready to be tested on other applications.

Full Text: PDF