Abstract
Currently, user interfaces displayed to viewers on their TVs look and behave similarly for all users. While sometimes it is possible to customise the user interface (UI) to a degree, users will rarely experience true customisation on a TV, mainly because that is difficult using a remote control or voice commands. Our research focuses on utilising machine learning to discover and interpret behavioural patterns and to adapt the UI accordingly. In this paper, we will share our solution for a truly adaptive UI, tailored to each viewer. This paper also showcases the machine learning engine and examines our behavioural mapping technique and the mathematical theory behind it.
Introduction
After years of researching digital television user experience (UX), we must accept that there is no simple and easy way for the viewer to customise the user interface. Even if there were a way for the consumer to achieve this, we have found a better, more powerful solution. What if the software could understand and learn what a specific viewer wants and adapt accordingly, as shown in Figure 1?
Personalisation of the digital television user interface has been identified as a crucial research topic, but most researchers to date have focused on pre-customised UI design and standardisation. The problem is that a standardised user interface will never provide an ideal user experience for each individual in the entire viewing/subscribing audience base because the needs and behaviours of each user can be vastly different from one another. A preschool child will have a very different mental model and thought process from an elderly couple watching TV together, or a young adult for example. Recent research in user experience increasingly emphasises viewers’ mental models, so the focus is shifted away from designing a solution towards understanding the user’s state of mind, and how we, or the service provider, can support those states.
It is obvious that the same user experience will not satisfy all users. This is why we need a way to ease customisation. If this is the case, how is it possible Figure 1 – AI customised UX that this problem has not yet been solved? The remote control itself is not ideal for the task. There have been suggestions for using gestures, pressure and breath as interaction mechanisms for interacting with a TV. Voice control seems to be a better approach, but it is still not ideal, easy to use or powerful enough. In recent years, user experience research has focused on using machine learning to create audience segments or personas. This still implies designing an experience for each persona manually, but it is a step in the direction of UI customisation.
No comments yet