I am a PhD student under supervision of Anne-Marie Kermarrec and Marin Bertier. I obtained a master degree in computer science at INSA de Rennes and IFSIC in 2007.

I am working for the Gossple project which aims at building a decentralized user-centered information system. We consider a folksonomy (item/tag information) in which each user has his own profile containing the information he contributed to the system. Using it, the user finds the other peers that belong to his interest community (clustering the network) and takes advantage of this collaborative knowledge to discover new items or tags. Current experimentation involve query-expansion and top-k recommendations and are deployed using traces from citeulike and delicious.

Some centralized architectures already provide such a personalized view of an information system, but they are limited by the computation power and the storage space required. In peer-to-peer system, the available resources grow with the size of the network. The challenge is no longer to be space or computation efficient, but to be able to identify and gather the relevant information which is distributed among all the peers in the system.

I also design a distributed collaborative filtering system in which each peer selects a bounded number of neighbors to get recommendations. I use traces from last.fm and edonkey to validate my algorithms.