Imperial College of Science, Technology & Medicine, UK
{y.arafa,p.charlton,a.mamdani}@ic.ac.uk
In this paper we examine the Personal Service Assistant (PSA) metaphor and analyse the requirements to support public information and service access. The PSA is still in the evaluation stage as researchers develop an understanding of the metaphor and the process for realising it. An emerging dimension to realising this metaphor is the desire to physically visualise the PSA in a life-like figure that not only exhibits intelligence in the manner it reacts to human interaction, but further portrays itself as an individual with character.
We discuss the experience gained from using the PSA, providing both a metaphor and infrastructure requirements as well as an evaluation of development. This paper concentrates on developing a base-system for supporting PSA development and considers how this prototype methodology can be expanded to support a more interesting set of features through defining a richer set of personality attributes and traits that serve to generically create a life-like character influenced by its emotions.
We aim to provide an emotional model for practical implementation of agents with synthetic consciousness. The implementation forms a test-bed to examine and evaluate how these emotions may influence agent personality and further assess how this may be reflected in agent action and visible gestures. We are looking at embedding emotional indicators in the content of the agent communication language and investigating the criteria and ways this may vary according to the context of the conversational acts and using this as a base for building agent's social and adaptive behaviour. Agent emotion and personality is a relatively new area and is, as yet, not well understood. Indeed, we do not fully understand how this works in humans as our emotional behaviour and attitudes are in many cases irrational and are not based on systematic patterns. Nevertheless, by defining an emotional model we hope to better understand this issue and better define the criteria and requirements for implementation, function and representation.