Affiliations: [a] Department of Computer Science, University of Hull, Hull, UK | [b] Department of Information and Control, Tongji University, Shanghai, China | [c] School of Engineering and Built Environment, Glasgow Caledonian University, Glasgow, UK
Correspondence:
[*]
Corresponding author: Ping Jiang, Department of Computer Science, University of Hull, Hull HU6 7RX, UK. Tel.: +44 1274 233940; Fax: +44 1274 236600; E-mail: [email protected]
Abstract: Virtual enterprises are currently characterised as having a fixed process model, fixed meta-data model and a fixed set of users. This paper proposes a behaviour network based approach to support dynamic and self-organised virtual organisations with less common knowledge and synchronising models, which are often the case in internet-based social interaction at its inception. The architecture is based on Maes' Behaviour Network method. Each agent assesses RSS feeds from others using a virtual sensor by calculating the semantic similarity between the incoming ambiguous information and its own interests. Each agent also contains a local behaviour network which uses an energy spreading mechanism to react to a combination of its own goals and state changes. An action-selector continuously looks for the best possible action to take, rather than carry out a model-based planning. A test-bed is developed on the JADE environment. A case study for new product development demonstrates the system self-organises a virtual organisation at its inception without a centralised model.
Keywords: Virtual enterprise, behaviour network, multi-agent systems, internet
