Canal-U

Collaborative Filtering: The Wisdom of the Internet

La sagesse collective : principes et mécanismes

Colloque des 22-23 mai 2008, organisé par l'Institut du Monde Contemporain du Collège de France, sous la direction du Professeur Jon Elster.

Intervention de Gloria Origgi.
“Internet not only provides the hugest information storage ever, but also the most sophisticated system of information retrieval people have dealt with in the history of culture. Information is evaluated and filtered before reaching its final destination. On Internet, people access content via the evaluations that other people have already left on that content. From search engines to reputational systems such as eBay, Internet is developing as a giant ranking system in which people's advices and preferences leave tracks that orients other people's advices and preferences: in many cases evaluation precedes or even replaces information: all that people look at on the Internet is tracks of opinions of others that will modify their judgement on a subject matter. In this paper, I provide an epistemological analysis of how the various techniques of collaborative evaluation work in Internet, to what extent are they reliable cues of quality and how the collective wisdom produced by the interaction of human preferences and search algorithms is achieved.
I will also try to argue that accessing evaluations and rankings is becoming a primary epistemological aim in knowledge search: I'll conclude by presenting a sketch of a “second-order” epistemology in which evaluative and ranking measures become the core ingredient in the extraction of information from a corpus of knowledge”.