Improving Adaptation Knowledge Discovery by Exploiting Negative Cases: First Experiment in a Boolean Setting

Tristan Gillard (1) , Jean Lieber (2) , Emmanuel Nauer (2)

1 BISCUIT - Bio-Inspired, Situated and Cellular Unconventional Information Technologies
2 ORPAILLEUR - Knowledge representation, reasonning

Case-based reasoning usually exploits positive source cases consisting in a source problem and its solution that is known to be a correct for the problem. The work presented in this paper addresses in addition of positive case exploitation, the exploitation of negative cases, i.e. problem-solution pairs where the solution is an incorrect answer to the problem, which can be acquired when the case-based reasoning (CBR) process fails. An originality of this work is that positive and negative cases are used both for adaptation knowledge (AK) discovery using closed itemsets built on variations between cases. Experiments show that exploiting negative cases in addition to positive ones improves the quality of the AK being extracted and, so, improves the results of the CBR system.

Mots clés: adaptation knowledge discovery; closed itemset extraction; case min-ing; negative cases; case-based reasoning

Domaines: Intelligence artificielle [cs.AI]

Tristan Gillard, Jean Lieber, Emmanuel Nauer. Improving Adaptation Knowledge Discovery by Exploiting Negative Cases: First Experiment in a Boolean Setting. ICCBR 2018 - 26th International Conference on Case-Based Reasoning, Jul 2018, Stockholm, Sweden. ⟨hal-01905077⟩