Research on CBR system based on data mining

Case based reasoning (CBR) is a popular problem solving methodology which solves a new problem by remembering previous similar situations and reusing knowledge from the solutions to these situations. Aiming at traditional CBR system's too much dependence upon experts or engineers, this paper introduces data mining technology into CBR system and GHSOM (Growing Hierarchical Self Organizing Map), an excellent data mining tool with ANN (artificial neural network) technology, is integrated with it. After principal features are selected from numerous initial features to represent a case, through GHSOM cases are organized and managed in case base and while case retrieval is conducted, new case is guided into corresponding sub-case base, which greatly raises system's accuracy and efficiency. At last, experiments are implemented to validate the effectiveness of the proposed methods by comparing the proposed methods with other recent researches.

► In this study we construct a model of case representation.
► We simplify case representation with the proposed feature selection method.
► We construct case base and conduct case management through GHSOM technique.
► We conduct case retrieval through GHSOM technique.