Editor’s Note: Dhia Mahjoub joined OpenDNS in early 2012 as a security researcher.
It’s been a great journey since I joined OpenDNS to work on exciting, cutting-edge research projects involving DNS, networks, security and big data. Given my background in research on graph algorithms, I was on the lookout for conferences about graphs in the city. My curiosity was rewarded when I discovered that GraphConnect was taking place in San Francisco at the Hyatt Regency on November 5-6. I registered immediately and counted down the days until the event.
The study of graphs has been around for over a century as a branch of discrete mathematics. With the advent of computers about 60 years ago, it became an important part of computer science. In the past decade, companies like Google, Facebook and Twitter respectively popularized the concepts of link/knowledge graph, social graph and interest graph. However the technology of graph databases, which was the topic of the conference, is still a nascent phenomena. Relational databases have been the de-facto technology to store and manipulate data in businesses. That is, until the NOSQL movement took the world by storm. Graph databases are the newcomer in this line of non-relational data stores, as graphs are both a natural and powerful way to model a plethora of problems we face in business and research.
The conference was successful in covering topics both technical and social. I noticed the presence of both big names like Twitter, Intuit, and Fujitsu and emerging startups such as FiftyThree and Squidoo that are embarking on the graph databases movement. It was also an opportunity for me to meet new people and share the awesome work we do and technology we deliver here at OpenDNS. In fact, the organizers gave all attendees badges with RFID tags and we were given the fun homework of approaching and getting to know as many people as possible as the tags will record our new connections.
My first favorite talk was the keynote by Professor James Fowler about “The Power of the Social Graph”. One of the interesting revelations from Dr. Fowler’s research is that smoking, obesity, and happiness spread within social networks between friends that are tightly related, since friends influence each other through emotions and behaviors. He also pointed out that these influence relations are not necessarily symmetrical and they differ based on gender. The other interesting discovery Dr. Fowler discussed is that genes affect voting behavior. The second talk I liked was titled “Intuit Payment Graph: A network built on payment transactions of customers and vendors” by G. Pillai. In this talk, Intuit presented a prototype of their Payment Graph whose data model is represented using Neo4j. Nodes are businesses or individuals and the links represent the transaction volumes and frequencies. The size of the model is a few tens of millions of nodes and links, and a few hundred millions of properties. They plan to use this system to connect consumers with small businesses, to create small business micro-communities, and to offer their customers referrals and recommendations. The next talk that caught my interest was “FluxGraph: A Time-machine for your graphs“ by D. Suvee. The talk discussed how graph databases can be augmented with the temporal notion so that we can travel the graph through time. Suvee discussed the useful application in retrospective medical research and presented the application they developed at Janssen Pharmaceuticals.
Overall, this conference was very useful as it gave me ideas on potential applications in our security research work at OpenDNS such as recommendations and temporal graph databases. I look forward to putting these ideas into practice and sharing the results with you soon.
Let’s not forget to mention that the organization of the conference, as well as lunch and dinner, were excellent, and the conference after party was so much fun :)!