Multi-participant Interaction in Multi-agent Naming Game

In this paper we analyse the influence of different interaction patterns on the behaviour of alignment processes in multi-agent Naming Game. We begin by introducing a meta-model of the Language Game that serves as a generalisation of the classical approach and facilitates better organisation and structuring of future research in the field. Further, we investigate the process against three interaction patterns (pair-wise, multi-speaker, and multi-hearer). The pair-wise interaction pattern involves a single speaker and a single hearer, participating in a single linguistic interaction; the multi-speaker and the multi-hearer interaction patterns assume multiple speaking and multiple hearing agents, respectively, involved in a single act of communication. All new types of interactions shape the performance of alignment processes that create the naming-convention. We show that the iteration-wise multi-participant patterns result in a visible improvement (increasing the number of speakers and the number of hearers results in a decrease of the number of interactions needed to reach a particular level of coherence), as compared to the classical pair-wise pattern. We show that an increase of the number of speakers and the number of hearers in multi-participant settings results in a decrease of the number of interactions, needed to reach a particular level of coherence.