Abstract
The Belief-Desire-Intention (BDI) agent model is a highly favoured agent development model known for its distinct abstraction between components, conceptual adaptability and flexibility in determining its actions. This determination is handled through a plan selection function which determines the most appropriate plan or action given the current state of the environment. Whilst it is conceptually easy to understand, the BDI platform remains guarded by a particularly steep learning curve, especially with regards to implementation and any required adaptation. Recent years have seen various forms of extensions and approaches to BDI agent models, including a model-driven approach based around the Extended Non-functional requirements framework. Non-functional requirements illustrate parts of a system which must be satisfied to an appropriate extent. These requirements remain indeterministic in their nature however, such that their satisfaction cannot be done directly. The model-driven approach within this paper uses components from this framework to formulate plans which are governed by their contribution to these requirements. This is done in an optimised manner to ensure the selected plan is optimal in regards to the systems overall attainment. To our knowledge, this is the first time an optimised approach has been used in relation to model-driven agent creation. This paper presents our optimised model-driven agent development approach, demonstrating its conversion from the initial extended non-functional requirements model into a completely optimised and functional agent. The approach is verified through experimental analysis.
Paper Type
Event
Optimal Requirements-Dependent Model-Driven Agent Development
The Belief-Desire-Intention (BDI) agent model is a highly favoured agent development model known for its distinct abstraction between components, conceptual adaptability and flexibility in determining its actions. This determination is handled through a plan selection function which determines the most appropriate plan or action given the current state of the environment. Whilst it is conceptually easy to understand, the BDI platform remains guarded by a particularly steep learning curve, especially with regards to implementation and any required adaptation. Recent years have seen various forms of extensions and approaches to BDI agent models, including a model-driven approach based around the Extended Non-functional requirements framework. Non-functional requirements illustrate parts of a system which must be satisfied to an appropriate extent. These requirements remain indeterministic in their nature however, such that their satisfaction cannot be done directly. The model-driven approach within this paper uses components from this framework to formulate plans which are governed by their contribution to these requirements. This is done in an optimised manner to ensure the selected plan is optimal in regards to the systems overall attainment. To our knowledge, this is the first time an optimised approach has been used in relation to model-driven agent creation. This paper presents our optimised model-driven agent development approach, demonstrating its conversion from the initial extended non-functional requirements model into a completely optimised and functional agent. The approach is verified through experimental analysis.
Recommended Citation
Goncalves, J. & Krishna, A. (2015). Optimal Requirements-Dependent Model-Driven Agent Development. In D. Vogel, X. Guo, C. Barry, M. Lang, H. Linger, & C. Schneider (Eds.), Information Systems Development: Transforming Healthcare through Information Systems (ISD2015 Proceedings). Hong Kong, SAR: Department of Information Systems. ISBN: 978-962-442-393-8. http://aisel.aisnet.org/isd2014/proceedings2015/MDDConcepts/10.