Using cultural and social beliefs in language games

Abstract

Agreement on word-object pairing in communication depends on the intensity of the beliefs that gradually emerge in a society of agents, on the condition that no one was born with embedded knowledge. The agents search and exchange ideas about unknown word-object pairings, until they meet a consensus about what the object should be named. A language game is a social process of finding agreement on word-object pairings through communication in a multi-agent system. In this paper, a technique is proposed to discover the association between a word and the agents’ beliefs on an object using self-organizing maps and a cultural algorithm in a multi-hearer environment. A conceptual space is implemented, which stores the agent’s beliefs in three dimensions, represented by colors. The technique was evaluated for a variety of scenarios using four significant measures: coherence, specificity, success rate, and word size. The results showed that with the proposed method social agents can reach agreement fast and that their communication is effective.