Goals

Given a word, find the words that have similar temporal profile
The user searches for a word in the UI and hits a "Find similar profiles" button, which would bring up a list of words with temporal profiles that look similar.

Given a curve, give list of words having the same temporal profile
The user manually enters a curve in the UI (by drawing it or specifying analytical functions) and she gets back a list of words with a matching temporal frequency.

Potential add-on: intersect with other database (countries, artists . . . ) in order to define ontologies (clustering of words)
Once we have an effective way of finding similar words we can detect for instance peaks in word use and match these to some open databases available on the web (dbpedia.org for instance). This could be presented on a nice timeline such as the one beside.


 

Potential add-on: merge all the metrics
We will come up with a lot of different distance metrics for comparing two temporal profiles, the idea of this add-on is to merge all these metrics together for instance in a user interface composed of sliders determining the weight of each metric.
 

Potential add-on: display data
One could imagine arranging the words along a circle in a specific order (for instance grouping them by some clustering criteria previously defined) and draw some lines between clusters according to similarities.