AbstractWhen designing large distributed systems, it is common to try and impose an or-ganisational structure, a structure with distinct roles, a deﬁned communication model, and social control aspects. Having an organisational structure allows for observing the system’s emerging overall behaviour, as well as controlling it. This seems more necessary when designing open systems where heterogeneous members are allowed to join and leave the system. Open systems especially need an organisational structure that is adaptive to changes, ideally a structure that is self-adaptive without the need of external intervention.
In the case of Multi Agent Systems (MASs), the building blocks of this self-adaptive organisational structure (the roles, communication model and a social order mechanism) need to be deﬁned in an unambiguous way that allows this structure to be applied onto software agents. Most of the existing models for designing Multi Agent Organisations make use of mental notions, such as power, belief, intention, obligation etc. These are notions used in every day language when describing an organisational structure, but can have an ambiguous mean-ing and semantics. This makes them diﬃcult to implement in a programming environment leaving their actual meaning open to the designer’s interpretation and leading to not truly open systems.
This thesis explores the possibility of bridging this gap between organisa-tional design and implementation by deﬁning the main building blocks of an organisational structure using a behaviourist approach. Inspired by the doctrine of Behaviourism, the thesis provides a descriptive framework for specifying an open organisation’s building blocks in behavioural terms.
|Date of Award||2018|
|Supervisor||Peter McBurney (Supervisor), Andrew Jones (Supervisor) & Michael Luck (Supervisor)|