algodiv - Information Diversity and Algorithmic Recommendation (ANR)The State, Political Norms and Political Conflicts
How is one to measure, check, and control the quality of choices made by the web’s algorithms whenever “they” decide to make visible this or that piece of information? What can researchers do in a context where concerns are becoming increasingly serious with regard to the place occupied by web calculators in the organization and in the hierarchy of digital information, as well as in the lives of private individuals? Can algorithms be “neutral”? What is the intent of those who conceive them? What ramifications does algorithmic guidance have on Internet users’ autonomy and freedom of choice?
This interdisciplinary project intends to make a scientific contribution to this political, economic, and cultural issue by clarifying the various meanings of the concept of information diversity, by proposing adapted metrics, by carrying out an empirical and experimental exploration of the combined effects of algorithmic guidance and behaviour of Internet users on this diversity, and by proposing alternatives to users. Closely linking computer scientists and researchers in the social sciences, Algodiv will focus on the challenges of information diversity in three separate configurations of algorithmic computing: structural authority measures built from the web's hypertext links; the social web’s affinity measures; and personalized navigation traces from a single body of data stemming from a partnership with the Groupe Le Monde. Computer and sociological surveys will be conducted in these three sectors to compare the rankings produced by platforms and digital practices – in some cases, experimental devices will be set up to more directly measure the link between the diversity offered and the diversity consumed.
You can find more detailed information on the project's website.
- Principal Investigator(s):
- Camille Roth