How can a historian evaluate the impact of on-line social networks on our perception of the historical past ? Trying to answer this question is not an easy task. It implies to be able to deal with huge amounts of data. For instance, since april 2014, more than two millions of tweets were collected, all related to the Centenary of the First World War. This presentation will explain how this corpus was collected, which kinds of statistical software are available for historians without statistical or coding skills and what are the limits of historical data-driven research when you are not a trained data-scientist.