Think Graph

A graph states that something is related to something else: people to people, molecules to molecules, habits to habits. Graph traversals are about paths: of ideas, time or viruses. Russian for “graph” is “граф,” which also means “earl.” In my adolescence, I thought that graph theory is about nobility, and despite my math education, a notion about nobility of graphs still sits somewhere on the back of mind. I pay attention to sprouting graph technologies not only as an analyst, but as an awed child and an ex-mathematician.

Earlier this week, I went to the first GraphConnect conference! I love inaugural conferences: they attract true enthusiasts — virtually anyone you get to talk to has an amazing insight or an amazing insider. Someone in the audience said, “So happy to not have to explain what a graph is.” I won’t explain it too, but will share my thoughts and takeaways. (My favorite book with explanations is Social Network Analysis for Startups, co-incidentally, by two Russian authors (I wonder, what they thought when they first heard about the graph theory.))

Somehow, it struck me at the conference that graph databases are noSQL. Maybe, as an analyst, I should classify them as such — but they could equally be noCobol, noFat or noNonsense. I prefer — yesFuture: up until today companies have been collecting data about entities: what do I know about this product, what parts it consists of, when this product was produced or where were we selling it. Organizations start thinking about relationships their data contains: what is the place of this product among other products, what its dependencies on the parts are, how do we divide the product revenue by a business unit? If it’s a customer, how is this customer related to others? Who is a customer of my customer? Who is a friend of my friend? I learned at #GarphConnect that targeting a network of friends is more effective than targeting an individual influencer.

Most speakers at GraphConnect were somewhat apologetic: graph databases don’t solve world peace but just a piece of a puzzle of the data-driven world. Don’t be apologetic – better evangelize! Explain, enlighten and elucidate what graphs are capable of. Do it in a small way and in the big way. Don’t be shy to say that graphs are a new and noble way of thinking about data. And yes, I tell my clients to start thinking what graphs can solve for them. I suggest, take the data you have and look at it as a graph — you will learn something new for sure (a pseudo-tweet).

Here are the highlights from GraphConnect on what some companies see in graphs:

Intuit wants to provide analytics and recommendations to small businesses under the motto “Big data for the Small Guy”. The main merit of a graph is in finding vendors of vendors.

Telenor in Norway uses graphs for… well… network analysis. Pretty obvious. And elegant. And 99.999% of other companies in the world are not doing it. Graph technology needs to mature not only technologically, but in people’s minds too.

Accenture is successful in applying graphs to logistics. There are two keys here: Accenture and logistics. One is the key to the implementation potential, another one is the key to the bunch of excellent use cases.

Cicso — Master Data Management! In the company with frequent acquisitions, graph is the best solution for ever changing organizational hierarchies, both for org mapping and for going back in time (as-is vs. as-was). Oh, yes, and real-time access control via permission resolution, especially when external partners are involved. Here is the link to details.

I have been thinking about graphs for MDM for awhile, and I’m glad, I am not alone. A thought is material for those who care about the same subject. In the stratum of graph thoughts, I caught ideas for personalized education, content management systems, stock trading, visualizations and playing what-if scenarios.

Coincidentally, on the day of the conference, I got a press-release, unrelated to GraphConnect, about Sherlock, a graph solution, discovering unknown relationships or patterns “hidden” in extremely large and complex bodies of information. So, Watson vs. Sherlock: Who will solve the data mysteries? Or, maybe, it would be Tom Sawyer? Or some neo-technology with noble ideas?