Formal description of a supervised learning algorithm for concept elicitation by cognitive robots

Abstract

Concept elicitation is centric for machine knowledge extraction and representation in cognitive robot learning. This paper presents a supervised methodology for machine concept elicitation from informal counterparts described in natural languages. The collective opinions of a given concept in ten selected dictionaries are quantitatively analyzed and formally represented according to the attribute-object-relation (OAR) pattern of formal concepts. The concept elicitation methodology for machine learning is aimed to deal with complex problems inherited in informal concepts of natural languages such as diversity, redundancy, ambiguity, inexplicit semantics, inconsistent attributes/objects, mixed synonyms, and fuzzy hyper-/hypo-concept relations. The system of formal concept elicitation is implemented by an algorithms in MATLAB for formal concept extraction and representation. Experiments on supervised machine learning for creating twenty primitive concepts reveal that a cognitive robot is able to learn synergized concepts in human knowledge in order to build its own cognitive knowledge base.