When NoSQL becomes noMapReduce : the need for large scale graph analysis frameworks

Cliquer ICI pour télécharger la présentation

Abstract : Map/Reduce (and its open source implementation Hadoop) is becoming the de facto standard for large-scale data processing, at least for easily parallelizable problems. But with the exponential growth in data, there is a need for a ubiquitous way to handle data. The graph model offers this, but graph computation is a domain where the Map/Reduce approach falls short. This presentation will give an overview of the existing solutions that try to tackle what is becoming one of the most important technological challenges: large-scale graph data analysis.  

We'll adopt the pragmatic point of view of an IT startup involved in social data analysis, while trying to give an overview of the forthcoming challenges, notably in the real-time analysis.

Biography : Julien Laugel is chief data scientist at MFG Labs, a startup company specialized in solving problems involving large scale data analysis, cofounded by mathematicians Jean-Michel Lasry, Pierre-Louis Lions, and net entrepreneur Henri Verdier.
Before that, he held various positions in R&D applied to the financial sector, including asset management, investment banking and data vendor companies.