Scaling Graph Computations with Online Algorithms

02:30 PM - 03:25 PM on August 15, 2015, Room 702

Andrew Kelleher

Audience level:

intermediate

Watch:

http://youtu.be/ddpRjHHPQrE

Description

There are a lot of ways to represent graphs, and a lot of interfaces and techniques for scaling them. We'll address how you might scale arbitrary operations on graphs by implementing a data stream that allows you to do your computations online.

Abstract

Working with medium-to-large scale graph data implies a lot of considerations. Optimizing for one use case usually has implications for another. Andrew Kelleher of Buzzfeed will illustrate some of the problems with querying graphs directly and run through the motivation for an online approach. Next he'll go over two algorithms for online computations on graphs given a few common use cases, and do a demo of them.