Sunday, December 13, 2009

Your cellphone can tell me what you did last summer

Our cellphones are nowadays an item, which we are very much accustomed to bringing along to every single place we go to. They are also devices that are very similar to little computers with multiple sensors and interaction modalities, they are equipped with GPS, Bluetooth, accelerometers, cameras, microphones,magnetometers,keyboards, and touch-sensitive displays,they also have great computation power and memory,graphics capabilities, and various communications capabilities.
All of these elements aside from providing a novel multi modal user interface experience give the means through which cellphones are a perfect device for tracing human activity. With all of the cellphone's sensors, one can obtain a collection of data, that is related with what a person did through out the day, then by using data-mining algorithms one can infer human relationships and behaviors, this is often refereed to as Reality Mining. The MIT Media Lab gives a far more formal definition of what Reality Mining is: "...R.M defines the collection of machine-sensed environmental data pertaining to human social behavior..."

The problem that is currently being faced is to understand exactly how the joint use of multiple modalities,like for example location and proximity to others, help understand a person’s routines. It is important to point out that many issues actually arise when one wishes to understand patterns in the life of an individual. It is not simple to automatically infer a person's activities as well as efficiently represent them . For example, having a stay home alone Thursday and a Thursday of Beer Hotness with friends at your place define entirely different social situations, yet they could be considered identical from the sole perspective of location. It is thus very important to have detailed descriptions of the activities done by a person for characterizing the users and their habits.

The big impact that reality mining has on us, is that it is able to create models of individual as well as group behavior from the recollected data, this could enable smart personal assistants, as well as monitoring of personal and community health.