RSTSM WG3 (Kuldar Taveter)

Start Date: 2012-02-16

End Date: 2012-03-01

Host Institution: Swinburne University of Technology

Host Country: Australia

Home Institution: Tallinn University of Technology

Home Country: Estonia


I plan to visit Prof Leon Sterling at Swinburne University of Technology in Melbourne, Australia.
Together with Prof Leon Sterling, we have written the monograph “The Art of Agent-Oriented
Modeling”, which was published by MIT Press in November of 2009 (Sterling & Taveter, 2009).
In the book, we proposed agent-oriented modeling as an approach for engineering
sociotechnical systems, which are systems that consist of both a social aspect, which may be a
subsystem, and a technical aspect. Agent-oriented modeling addresses sociotechnical systems
from three balanced and interrelated viewpoint aspects – interaction, information, and
behavior – each with its particular set of models, which is determined by a particular design
methodology used. In addition to representing for each model the vertical viewpoint aspect of
interaction, information, or behavior, each model can be mapped to the abstraction layer of
analysis, design, or platform-specific design.

The purpose of the visit is to initiate a joint project in applying agent-oriented modeling for
designing innovative products as socio-technical systems. Products that can and should be
engineered this way include buildings, cars, domestic applications, industrial automation
systems, and enterprise information systems. Because of the scope of the COST Action
“Agreement Technologies”, the project will put special emphasis on designing socially-oriented
information systems where naturally distributed social structures are supported by software
applications running on many devices rather than by one centralized application running on a
server. This is particularly relevant now when we have recently seen remarkable advances in
networked mobile “smart” devices, such as iPhones, Androids, and iPads. Software applications
that run on individual devices assisting and representing their principals are called agents.
Earlier collaboration between Prof Sterling and me in this area has involved designing a
prototypical agent-based system for selecting, playing and sharing music based on human
activities and emotions (Luo, Sterling & Taveter, 2007). As in such systems the overall solution
for a particular principal is delivered through finding an agreement between many agents
representing their principals, the proposed project is very relevant for the COST Action IC0801
“Agreement Technologies”, in which both Tallinn University of Technology and Swinburne
University of Technology participate. The areas where the resulting agent-based information
systems can help are regulation (e.g., banking), allocation of scarce resources (e.g., electric
power, parking spaces, and emergency care), distributed situation assessment (e.g., traffic
jams), and decentralized decision-making (e.g., finding a healthcare provider), which represent
the four kinds of problems that societies confront. For example, Taveter, Du & Huhns (2012)
describe the design of an online system where customers could post the prices paid for
groceries (this could be automated by scanning barcode labels of the products and later on by
querying the RFID tags of the products) and where a prospective shopper could enter a grocery
list and obtain a pointer to the store with the lowest total price. This would enable comparison
shopping for groceries and would render the customer-to-store interactions fairer. It would also
encourage stores to offer their true prices to avoid driving away potential customers. However,
the effort required from the customers would be substantial. To make the effort reasonable
and manageable, each customer could benefit from an agent that represented his/her interests
and interacted with the agents of the other customers and, possibly, with store agents.