Incertaine

Participants: Juliane de Moerloze, Quentin Lacroix

This project questioned the contemporary faith people have in the application of machine learning algorithms, by subverting models of logic and statistics.

A very limited amount of individuals received a series of very specific questions about the interpretation of the adjective 'corrélatif'. Next, they applied two different methods of analysis of these data, on the one hand interviewing the individuals, and text mining on the other. The results turned out to be uncomparable. Moreover, a logic based exercise showed that langage is such a subjective material that it is capable of diverting any probability algorithm, because it is literally impossible to represent a word by a dot. The context of a word is so meaningful for the interpretation of a word, that one should be able to insert this complexity in the dot of a graph.

What does one measure while analysing text? And are we really measuring the same thing? By this poetic approach the project shows how we permanently live with a shadowy margin of doubt, incertitudes and multiple interpretations, ans how Big Data makes us sometimes believe there is only one truth.