UCSD Cognitive Science Honors Project
Overview
For my honors project I delved into methods of organizing large amounts of data. In particular, I explored the data generated by life logging techonologies. With the increased production of information, and the ability to cheaply store it, more and more data is continually being captured. I explored a method known as segmentation to see what the capability would be to automaticailly categorize it.
Study
I used three devices to capture data: the Microsoft Sensecam which recorded pictures, the live scribe pen which recorded drawings and writting, and the ActivityTrailsLite which recorded computer screens. Each participant broke their recording into segments, meaningful chunks. An example would be a student breaking their visual recording into going to class, being in class, and then leaving class. The participant would then segment two other participants data.
Results
The easiest data to segment was the written data produced by the smart pen. The other datasets varied depending on the complexity of the activity being recorded and whether there was context needed to help identify what was occuring. Overall, it looked like an algorithm may be able to get some useful chunks out of it, but it would likely require quite a bit of user modification to get a useful set for complex sets.