Human behaviour is just too random to be captured in a precise manner by contextualisation software. However, we can get round the problem by turning to a manual keywords system.
There are plenty of applications that enable automatic identification of the user context (photo tags, etc). But this identification remains partial and often imprecise because of the difficulty of identifying the activity of the user in question. Now Matthias Böhmer, Antonio Kruger (researchers at Germany’s Centre for Artificial Intelligence), and Gernot Bauer suggest going back to the source – i.e. calling on Internet users themselves to refine the context. "Nothing is more efficient than collecting information at its source – which in this case means from the consumer himself", they underline in their research paper.
Two different users tend to use the same keywords for the same situation
In fact, according to the researchers, people with related goals will tend to use the same keywords. They became aware of this when studying the various types of "context tags" which users submitted to describe their situation. By contrast, entirely distinct contexts will throw up distinct indexation. Basically, the “folksonomies” of different users seem to widely accord with one another. From a practical point of view, this finding has opened the way for the development of a very simple but remarkably accurate application.
A test application designed to demonstrate the principle’s potential
By placingthem on an interactive chart, users can offer various services - such as a café, a shoe repair service, etc - by linking them to specific keywords. As the study shows, a user searching for a particular service will tend to search using the same words as the supplier of the service. Any potential ambiguity will disappear, and the search will be made far easier. According to the three researchers, this application provides just a mere glimpse of the possibilities offered by this contextualisation method, - a method which lends itself to continuous change in line with user needs.