On Sunday mornings I generally try to grab some quiet time so I can catch up on interesting articles I may have missed during the week. Occasionally I read something that provokes a blog post. This is one of those days and since it’s absolutely miserable out there (the Irish equivalent of a monsoon) I figured I might as well put pen to paper now. The article in question is an InfoWorld post from James Kobelius (@jameskobielus) entitled Graph analysis will make big data even bigger. It’s a nice succinct article that talks to the potential value of social graph analytics and also the challenges of storing and analyzing these massive graphs.
Over the last while there have been lots of discussion about the over-hype of big data and the fact that small data is often enough and while I would agree that this is often the case, the graph is one of those areas where small data gets very large very quickly. This is most especially the case if you believe, as I do, that the Social Graph is just the tip of the iceberg when it comes to doing people-centric analysis.
Today when folks think about social graphs they tend to think about direct interactions between people within a social network. However, when you think about analytics these social interactions are only a small part of the picture and what you really need is the ability to integrate the social graph, knowledge graph, business graph (derived from analyzing business activity), and ultimately the web of data.
For example; if we take a very simple marketing scenario “how do I know who to target with certain marketing messages”, then just imagine if you knew:
- which people were interested in your product (social graph).
- which were already customers (business graph).
- which would likely have the most use for your product (knowledge graph).
- the optimal channel, time, location, device, and application through which to share the message (web of data).
I’ve been struggling in my own work to come up with a definition that adequately describes the hybrid nature of this new graph. Some clients like to call it “the business network”, while others prefer to stay with “the social network” moniker. Whatever we end up calling the graphs just remember that the value of social is only truly realized (at least for analytics) when it gets integrated with non-social data.