Scalable Facility Location for Massive Graphs on Pregel-like Systems

Kiran Garimella · Mon, 07.12.2015


This work is one of my first major projects in my PhD, which deals with creating algorithms for very large datasets using recent advances in commodity computing clusters (like Hadoop). We designed and implemented this algorithm on a large cluster containing hundreds of machines and demonstrated the effectiveness of our algorithm. This work was in collaboration with Aris Gionis (Aalto University, my PhD supervisor), Gianmarco De Francisci Morales (Yahoo Research, at the time of publication), and Mauro Sozio (Telecom ParisTech). The conference (CIKM) is one of the best in the areas of data mining, and it was a great experience interacting with a lot of researchers from all over the world. There were keynotes by renowned speakers from Google, Microsoft Research and University of Melbourne. There were also a lot of interesting talks in the area of graph mining and social network analysis, which are relevant to my thesis.

I wrote a short blog post on our group blog describing the paper, and my experience at the conference.