Virtual Personal Assistants - A Different Approach to Supporting the End User

My research is focused on developing a model of a virtual personal assistant, a technology that interacts with one individual and develops a personal relationship to provide more contextually relevant outputs to the user. We are exploring similar technologies that exhibit different aspects of the model of technology we are describing to evaluate the usefulness of each component individually, although there is no reason why in the future they couldn’t all be combined into a new technology. What these similar technologies lack is the personal feedback loop between the end user and the technology, where the personal assistant adapts to the end user and the user learns from the systems outputs. Supporting technologies are becoming more involved in people’s lives, both on a personal and professional level. This is in part due to the rise of new cutting edge technologies that allow these supporting systems to attempt to be useful in different aspects of a user’s day to day life. The category of supporting technologies we are interested in are systems that specifically support the end user [1]. Examples of these systems typically have the ability to build metadata from a combination of inputs and stored data to be able to modify their outputs. An example of this can be seen with IBM’s Watson, a system that can provide a medical diagnosis of a patient based upon its understanding of the symptoms [2]. Watson learns through analysing medical journals and comparing that to what it has already learnt from in order to provide the most logical diagnosis [3]. A number of modern supporting systems are utilizing learning capabilities to build a relationship with the user in an effort to provide more refined and useful outputs. A recent example of this is Apple’s Siri, which provides support to its user in a casual context with enhancements based upon its knowledge of the user [4]. Siri learns from its interactions with the user and categorises the user with other users who Proceedings of STPIS'17