Laurent Massoulié Grand Prix Cino Del Duca
On 7 June, Laurent Massoulié, head of the Infine team at the Inria Saclay – Île-de-France Research Centre and Director of the Microsoft Research - Inria Joint Centre, received the Grand Prix Scientifique Cino Del Duca 2017 , awarded by the French Academy of Sciences. He has won this prestigious science prize for his project on Social Information Networks.
What is your research about?
I work on detecting communities in social networks. Based on how users interact, we can define one or more statistical user profiles. Profiling enables us to recommend relevant content likely to be of interest to the user or to put them touch with other social network users.
More specifically, my research focuses on developing algorithms which produce a relevant profile at a critical level of data quality below which they are too poor for information profiling. Below that level, there is not enough data, and what there is cannot be used to produce relevant statistical profiling.
What have you and your team discovered?
My team and I have developed what are known as spectral algorithms, which can be used for information profiling of network users up to a critical level of noise. This transition point is the threshold below which no known algorithm can produce such profiles. This phenomenon had been predicted before by researchers in Statistical Physics. We have simply confirmed it.
What concrete applications are there for your discoveries?
For online social networking, there are intermediation platforms, in particular, platforms for users to ask and answer questions, such as Stackoverflow and Quora. Answers are rated by other users of the system. If an answer is unsatisfactory, an expert can give their version to go deeper into the question. Such networking works extremely well in IT, for example. Development in this area is strong, with groups for all sorts of everyday activities, such as gardening, etc.
Another example of an intermediation platform is to connect jobseekers with job offers, such as Upwork in California. Here too there is a need for statistical profiling to match users with job offers.
These are issues that you are working on at the Microsoft Research - Inria Joint Centre, of which you are the Director.
Yes, Microsoft is interested in these issues since it deploys intermediation platforms. It is in all our interests to identify effective algorithms. Microsoft can also help to put us in contact with companies that have attractive innovation opportunities.
What does winning the 2017 Grand Prix Scientifique Cino Del Duca 2017 mean to you? How do you plan to use it?
It is really fantastic, and I am so grateful to the Cino Del Duca Foundation for the award. The prize comes with a sizeable award, worth €200,000, which means we can take on a post-doctoral researcher and organise conferences on the subject. Together with my three colleagues (Milan Vojnovic from the London School of Economics, Alexandre Proutière from the KTH in Stockholm and Charles Bordenave from the Toulouse Mathematics Institute, CNRS), we will supervise this post-doctoral researcher and hope to make swifter progress in achieving our objectives.
What are your objectives?
We have two major algorithm challenges at the moment: community detection in intermediation platforms and iterative recommendations on these platforms. If an expert replies to a question and their reply is deemed unsatisfactory, the platform can recommend another expert. We can then develop an adaptive recommendation algorithm which will enhance the performance of the platform. We are hoping that we can really make use of this aspect in the future.
Biography of Laurent Massoulié
After graduating from France's École Polytechnique, Laurent Massoulié completed his PhD in Applied Probability at Supélec in 1995. He then worked for many years at industrial research laboratories (France Télécom R&D, Microsoft in Cambridge and Thomson, now called Technicolor) before joining Inria in 2012. He began as leader of the Infine research team at the Saclay – Île-de-France Research Centre and was appointed as Director of the Microsoft Research - Inria Joint Centre.
After working on data flow management on the Internet and peer-to-peer systems, his research currently focuses on the statistical profiling of online social network users and the information disseminated in such networks.