New models to improve the reliability of virtual organisations
Models of virtual organisations that operate within reliable and trustworthy parameters are being developed by researchers at the University of Southampton.
A virtual organisation is one whose members are geographically apart (usually working via networked computer applications) while appearing to others to be a single, unified organisation with a real physical location.
According to Professor Michael Luck of the School of Electronics and Computer Science (ECS), as the market for virtual organisations grows and an increasing number of companies are represented by computerised agents acting on their behalf, there is a greater need to ensure that these agents behave responsibly.
In a project which aims to monitor the performance of the members of a virtual organization in terms of their trustworthiness, quality of service and contract compliance, Professor Luck and his team at Southampton have worked with Cardiff University, the University of Aberdeen and British Telecom on CONOISE-G (Grid-enabled Constraint-Oriented Negotiation in an Open Information Services Environment).
‘Creating and then effectively managing a virtual organisation in a dynamic environment poses significant research challenges,’ commented Professor Luck. ‘We need to draw inspiration and examples from human societies and apply them to computerised societies. Selecting different suppliers on the basis of reputation, using information from others to decide who to trust, and discounting information from unreliable sources in making judgements are all actions that need to take place in the interactions of computerized societies as much as in our normal daily lives.’
In seeking to address some of these challenges, the researchers have developed a system for the dynamic formation and operation of virtual organizations, drawing on scenarios such as that of an individual visiting London for the 2012 Olympic Games who requires a PDA to access various multimedia services.
They are currently in the process of implementing a prototype system which looks at issues such as trust and reputation, standardising communication between agents, and policing within a virtual organization, so that the impact of behaviour such as non-delivery of services by an agent is minimised.
Professor Luck commented: ‘The trustworthiness and reputation of agents are significant issues, especially in the context of virtual organizations in which the agents must rely on each other to ensure coherent and effective behaviour.
‘Only limited work has been carried out in this area so far, with the majority of developers adopting the stance of complete trust. This, however, avoids the complex issues which are crucial for the reliability and dependability of these systems and which our research aims to address directly.’
CONOISE-G is due to be completed this month.