Two concrete application scenarios will be developed, each of which enhances products that are already on the market. The first scenario, `eShowroom' explores online shopping. Users will create consumer agents in their own image, and these agents will go off into cyberspace to negotiate with producer agents, after which they come back to their creator to explain highlights of what they have found. Crucially, they combine words and body language to report on their findings, making it easy for the user to choose a product. The main benefit for the user lies in the fact that agents can sift through huge amounts of data, reporting only on what is relevant.
The second scenario, `socialite' uses the same technology for a different purpose. As in the eShowroom, users create agents which they send into a virtual world where they will meet other agents. This time, however, the goal is entertainment rather than commerce: the goal is not to buy a product but to enjoy seeing the agents interact, and to be able to influence their behaviour by changing their personality traits. To make this possible, the system makes visible these interactions by means of a generated `cartoon'.
There are believed to be growing opportunities for the deployment of agents of this kind, and a growing body of work in Information Technology addresses the enormous technological challenges involved in creating them. The main task for ITRI will be to put abstract interactions into words, and to do this is as natural a way as possible. For example, in the eShowroom scenario, when an agent reports back on a second-hand car that costs £3000, it might not always be natural to just say `The car costs £3000'. If £3000 is cheap for a car of this kind (a 1995 Mercedes, for example) and within the price range specified by the user, then the agent might express his enthusiasm by saying `Did you notice that this car, a Mercedes for heavens sake, costs only £3000 pounds?!' - with appropriate intonation and body language of course. ITRI will be responsible for language generation, which will allow us to build on our extensive experience in this area, and to extend the range of expressions that our generators can generate. The other academic sites will focus on psychological modeling of agants and on speech synthesis. Sysis is responsible for graphics generation. Both Freeserve and Sysis will be responsible for the exploration of commercial opportunities.
Maintained by Paul Piwek
Last Modified: 16 November 2001
© Information Technology Research Institute at the University of Brighton