Case-Based Reasoning Systems
This book reports on a set of recently implemented intelligent systems, having the case-based reasoning (CBR) methodology as their core. The selected works witness the heterogeneity of the domains in which CBR can be exploited, but also reveal some common
- PDF / 229,224 Bytes
- 6 Pages / 439.37 x 666.142 pts Page_size
- 76 Downloads / 253 Views
Case-Based Reasoning Systems Stefania Montani and Lakhmi C. Jain
Abstract This book reports on a set of recently implemented intelligent systems, having the case-based reasoning (CBR) methodology as their core. The selected works witness the heterogeneity of the domains in which CBR can be exploited, but also reveal some common directions that are clearly emerging in this specific research area. The present chapter provides a brief introduction to CBR, for readers unfamiliar with the topic. It then summarizes the main research contributions that will be presented in depth in the following chapters of this book.
1 Introduction Case-based reasoning (CBR) [1, 10] is an Artificial Intelligence (AI) technique meant to provide automatic reasoning capabilities, while allowing continuous learning, in advanced decision support systems. Specifically, CBR exploits the experiential knowledge collected on previously encountered and solved situations, which are known as cases. The reasoning process can be summarized using the following four basic steps. These are known as the CBR cycle (Fig. 1), or as the four “res” [1]. The procedure is to: (1) retrieve the most similar case(s), with respect to the current input situation, contained in the case repository, which is known as the case base; (2) reuse them, or more precisely their solutions, in order to solve the new problem; some adaptation may be required at this stage; S. Montani DISIT, Computer Science Institute, Universita’ del Piemonte Orientale, Viale Michel 11, 15121 Alessandria, Italy e-mail: [email protected] L. C. Jain (B) Faculty of Education, Science, Technology & Mathematics, University of Canberra, Canberra, ACT 2601, Australia e-mail: [email protected] S. Montani and L. C. Jain (eds.), Successful Case-based Reasoning Applications-2, Studies in Computational Intelligence 494, DOI: 10.1007/978-3-642-38736-4_1, © Springer-Verlag Berlin Heidelberg 2014
1
2
S. Montani and L. C. Jain
Fig. 1 The case-based reasoning cycle
(3) revise the proposed new solution (if necessary), to build a new case; (4) retain the new case for possible future problem solving. In many application domains it is also common to find CBR tools which are able to extract relevant knowledge, but leave the user the responsibility of providing an interpretation and of producing the final decision: steps reuse and revise are not implemented. In fact even retrieval alone may be able to significantly support the reasoning task [16]. As previously stated, CBR not only supports reasoning, but combines problem solving with continuous learning. Indeed, CBR relies on experiential knowledge, in the form of past problem/solution patterns. It does not aim to generalize from past examples, and to learn general rules/models from them, as it happens in other AI reasoning techniques. Indeed, CBR keeps and exploits the specific instances of problems which have been collected in the past (almost) “as they are”. By using CBR, the difficulties of knowledge acquisition are therefore often lessened,
Data Loading...
