In this document we show how combining case-based reasoning and knowledge-discovering techniques can help significantly to acquire the knowledge contained in a case base. In our particular application, the case base describes the forest fire emergencies recorded in the past in a department of the southern France. We have called the system CBET (Case-Base Exploration Tool). CBET is a comprehensive system for case base manipulation, browsing and enquiring. The major focus of the system is retrieval for learning, and we have addressed this problem by using both CBR and KDD techniques. So for example retrieval based on a Nearest neighbor algorithm, a standard technique in CBR [2], can be supported by a selection of relevant features, performed with statistical and information theory algorithms [21]. CBET supports the user from the definition of the data structures, to the modification and maintenance of the case base. CBET provides a set of functions for: retrieving cases that match a partial definition; evaluating the relevance of features and group of features in forecasting other feature values; clustering cases according to distances oriented to the application domain; plotting bogh symbolic and numeric data
A Case Base Exploration Tool
Avesani, Paolo;Perini, Anna;Ricci, Francesco
1997-01-01
Abstract
In this document we show how combining case-based reasoning and knowledge-discovering techniques can help significantly to acquire the knowledge contained in a case base. In our particular application, the case base describes the forest fire emergencies recorded in the past in a department of the southern France. We have called the system CBET (Case-Base Exploration Tool). CBET is a comprehensive system for case base manipulation, browsing and enquiring. The major focus of the system is retrieval for learning, and we have addressed this problem by using both CBR and KDD techniques. So for example retrieval based on a Nearest neighbor algorithm, a standard technique in CBR [2], can be supported by a selection of relevant features, performed with statistical and information theory algorithms [21]. CBET supports the user from the definition of the data structures, to the modification and maintenance of the case base. CBET provides a set of functions for: retrieving cases that match a partial definition; evaluating the relevance of features and group of features in forecasting other feature values; clustering cases according to distances oriented to the application domain; plotting bogh symbolic and numeric dataI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
