From a mobile phone user’s perspective, some degree of intelligent widget control should be provided as one goes about his/her daily activities. For instance, a user driving to work will more likely use the phone in “hands-free” mode for voice communications instead of SMS or email. A student who normally uses the phone for entertainment and social interaction will likely not use it during curriculum time. It is also less likely that an employee would want to invoke a game application or view a movie at the work place than when travelling via public transport or resting at home in the evening. Similarly, a mobile business executive or tourist may place more emphasis on GPS and location-based applications and functionality than an office-bound employee, albeit using different categories of services.

We ask the following questions:

• Does a user need all or even most of a mobile phone’s capability?

• Does a user need the same subset of applications and functionality all the time?

• Can a mobile phone be made to learn and recognize a user’s context?

It is quite apparent that different subsets of a mobile phone’s capabilities appeal to different users depending on roles/interests. On a regular basis, many users also make use of specific applications and functionality in a fairly deterministic pattern depending on context. A typical mobile phone user’s context may be defined in terms of usage pattern, date, time of day, and location as a basis. With the aid of suitable sensor inputs, additional contextual information may be gleaned e.g. how far has the user moved from the previous location, how fast is the user moving now, heart rate (stress level) etc. However, at the time of this paper, there is no reported work that offers a learning engine with dynamic context-aware reconfiguration of a mobile phone interface. As a consequence, mobile phone users would have to constantly navigate through and manually reconfigure a complex and confusing set of excess widgets that they either do not need or no longer use.