PAIS Lab seminar: Knowledge and beliefs of agents in the cloud

At the session of our seminar, Nikolay Shilov (senior researcher at A.P. Ershov Institute of Informatics Systems) delivered a report "Knowledge and beliefs of agents in the cloud".

Abstract:
Distributed system is a group of decentralized interacting coputational nodes. Distributed algorithm is a communication protocol in a distributed system which turns a decentralized group into a team working together to solve certain problems. Multi-agent system is a distributed system consisting of autonomous reactive agents, whose internal states can be characterized in terms of beliefs, desires, and intentions. Multi-agent algorithm is a distributed algorithm for a multi-agent system.

Many problems, which are solved by multi-agent algorithms, can be considered as examples of social software problems .Nowadays, a lot of social demands and procedures have a very clearly described algorithmic nature. Therefore, these requirements can be written as (semi) formal specifications and procedures - program (for any programming language or pseudo), then the properties of these procedures can be studied by the methods of program analysis and verification. Results of analysis and verification can be interpreted as socially significant terms. Although social software spoke gained populatiy only in XXI-century, the first example of the application of this paradigm can be considered the research of H. Steinhaus, B. Knaster, and S. Banach on task of cutting the cake.

In this report, I will consider an example of the use of multi-agent algorithms in the context of social software engineering, which can be associated with the problem of cutting the cake or the problem of rational agents in the market. "Cake" is in this case represent a resource center or a cloud of available resources for which agents compete.