The resolution of symbol meaning in multi-agent systems

This research assumes a language game model of interaction between autonomous agents and studies three meta-models of alignment, i.e. models based on the elimination strategy, frequency strategy, and proportional strategy, implemented in two novel models of semiosis, individual - with predefined semantics - and population - without predefined semantics. The introduced models are analysed using a developed simulation framework and a defined set of basic measures that resemble the dynamic character of the alignment process. The obtained results explain a number of theoretical and practical issues related with the process of semiosis in multi-agent systems. In particular, they establish a basic set of criteria to asses a specific implementation of the communication system, understood as a set of components (model of environment, interaction, population, agent and alignment strategy) applied to a particular model of semiosis. Research's thesis: The characteristics of the external world, interaction, population and agents affect the behaviour of the system undergoing a dynamic process of symbols' meaning resolution. There exists a mixture of the characteristics that allows a particular language alignment strategy to result in a formulation of coherent substance of symbols within the interacting population.