An approach to determination of representative set of exemplars for basic-level category mapping in cognitive agents

• Published in: Procedia computer science Vol. 207; pp. 4237 - 4247
• Main Authors: Lorkiewicz, Wojciech, Mulka, Mariusz, Katarzyniak, Radosław
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 2022
• Subjects: basic-level categories; cognitive agent; concept mapping; concept mapping; cognitive agent; basic-level categories
• ISSN: 1877-0509; 1877-0509
• DOI: 10.1016/j.procs.2022.09.487

Cognitive agents should be equipped with computational methods enabling autonomous formation and management of their internal cognitive structures. Moreover, due to social nature of multiagent systems a dedicated subsystem is required to allow for establishing common structures allowing for successful interaction between agents. Current implementation tend to use shared and predefined system-wide ontologies, thus limiting the fexibility and autonomy of individual agents. In this research, we focus on an approach that can remove this dependency by allowing the agents to maintain individual internal ontologies and map them to internal ontologies exploited by other agents. In particular, we propose a general strategy in which agent engages in an interplay that enables determination of certain types of relation between concepts. Further, we focus on one of the initial steps of this strategy, namely, determination of the representative set of exemplars for basic-level categories, which is utilised during the interplay and serves as the baseline to identify the mapping.