Adaptation Knowledge Discovery Using Positive and Negative Cases

Case-based reasoning usually exploits positive source cases, each of them consisting in a problem and correct solution to this problem. Now, the general issue exploiting also negative cases—i.e., problem-solution pairs where answers incorrectly problem—can be raised. Indeed, such cases are “naturally” generated by CBR system as long it sometimes proposes incorrect solutions. This paper aims at addressing for adaptation knowledge (AK) discovery: how can used purpose. The idea is that propose rules filter out some these rules. In preliminary work, kind AK discovery has been applied using frequent closed itemset (FCI) extraction on variations within case base tested toy Boolean use case, with promising first results. resumes study evaluates 4 benchmarks, which confirms benefit discovery. involves adjustments data preparation rule filtering, particular because FCI works only features, hence methodology lessons learned cases.