CEC-Model: A New Competence Model for CBR Systems Based on the Belief Function Theory

Abstract

The high influence of case bases quality on Case-Based Reasoning success gives birth to an important study on cases competence for problems resolution. The competence of a case base (CB), which presents the range of problems that it can successfully solve, depends on various factors such as the CB size and density. Besides, it is not obvious to specify the exactly relationship between the individual and the overall cases competence. Hence, numerous Competence Models have been proposed to evaluate CBs and predict their actual coverage and competence on problem-solving. However, to the best of our knowledge, all of them are totally neglecting the uncertain aspect of information which is widely presented in cases since they involve real world situations. Therefore, this paper presents a new competence model called CEC-Model (Coverage & Evidential Clustering based Model) which manages uncertainty during both of cases clustering and similarity measurement using a powerful tool called the belief function theory.

An evidential integrated method for maintaining case base and vocabulary containers within CBR systems

“…Besides, we find other CBM policies [30,31,32,33] that apply maintenance operations according to some evaluation criteria such that (1) the performance which is quantified by the time spent to solve a target problem, and (2) the competence which represents the range of problems that the CB can successfully solve [26,29].…”

“…The idea behind dividing a CB into some case types has been used in several works such in [27,34,36,38,40,41]. This categorization is generally based on two fundamental concepts related to the CB's competence called Case reachability and Case coverage [29]. Hence, we detect our four types of cases based on the following definitions [40]: Definition 4.1.…”

EXPLOITING PRIOR KNOWLEDGE FOR EVALUATING CBR SYSTEMS USING A COVERAGE & CONSTRAINED EVIDENTIAL CLUSTERING BASED MODEL

“…| | − 1 We may then calculate the density of the group G by averaging all its local densities, such that: Last but not least, we aim to estimate the coverage of each group. Similarly to (Smyth and McKenna, 1998) and (Ben Ayed et al, 2018a), we are interested to estimate them according to some characteristics, as follows:…”

“…However, it is essential, on the one hand, to handle uncertainty and imperfection attached to cases since they reflect real-world situations. In fact, we note, in the literature, a high interest in evaluating the competence of CBR systems' CBs (Smyth and McKenna, 1998;Smiti and Elouedi, 2013), where some of them prove the necessity of managing uncertainty within such systems (Ben Ayed et al, 2018a;Weber, 2006). On the other hand, there is no research, to the best of our knowledge, that provide the ability to available prior knowledge to aid learning and to lead a better evaluation.…”

“…After 3solving problems, new cases are retained within the CB at the aim to improve systems learning capabilities (Aamodt and Plaza, 1994). Since CBR systems have known a widespread utility in several domains, and since their knowledge containers present their success key factors, a wide range of research intend to assess their performance and competence (Smyth and Keane, 1995;Smyth and McKenna, 1998;Smiti and Elouedi, 2013;Ben Ayed et al, 2018a). Due to its distinguished reasoning, different concepts related to CBR systems evaluation have been proposed, such that (1) the Coverage which defines the space of problems that one case or a group of cases are able to successfully solve, and (2) the Reachability which presents the space of cases that is able to solve some target problem(s).…”