Data Awareness; Operational Intelligence - How IT Can Add Value
By Mike Matchett - Taneja Group
This is a big title deserving a full blown report, maybe even evolving market landscape research, but here are a few teasers to get your juices flowing -
1. IT should be constantly striving to provide more value and return back to the business
2. Data is too valuable to just persist, protect, backup, and archive. Faster competition, tons of new data, and the Internet of All Things mean IT has a huge opportunity to step up to the plate and deliver better ways for the business to harvest value from what in the past would simply be just data.
What if IT could deploy a "data service" (we are all becoming more like service providers, right?) that auto-magically on ingest turned all kinds of data into a big coherent linked data set? A giant Facebook for the business and all it's data entitites? With that service, business users and operationally focused apps could immediately query across all data (think data lake and externally linked in data too), not looking for specific records as in SQL or certain terms as in search (although these are powerful paradigms too), but to find inherent patterns and relationships between business objects and "entities".
I'm talking about a kind of operational intelligence mining based on graph theory. Which is not that mysterious when you simply think of it as a network of "nodes" connected by "links". The cool emerging part here is that there are new solutions IT can deploy that will suck in otherwise silo'd data and build networks of information up relatively automatically, and then basically make them serve as massive business oriented sources of intelligence. In common usage, they can work like "recommendation engines", or help solve that "six degrees from Kevin Bacon" problem (nested friend of a friend queries).
You might be most familiar with Facebook's "social graph" or LinkedIn's "economic graph", but imagine if all your structured and unstructured (processed on ingest) business data were somehow all in some kind of business oriented Facebook or LinkedIn for your users? Your users could "traverse" that graph with relative Facebook-like ease, and everyone could be on the same up-to-date, immediately available data view.
And by the way, this graph thinking approach implies that there is no target schema work required ahead of time, no massive DW effort, and no limit to the kind or type of data that gets linked/associated (or the type of link/association either). And the opportunity and value of it grows with its growing size!
Check out a pair of hot companies powered by variants of this approach that we've heard from this week - Reltio Cloud and Saffron. Reltio's practical solution melds MDM and graph theory together to help IT master and serve up its various data sets linked together in a "commercial graph" with immediate business user friendly application. Saffron's "Natural Intelligence Platform" is more autonomic in that it learns only from what's in the data ingested (no external training). It's aimed less at cleansing/mastering various data on ingest and more on building up a deeper "associative memory" of what it has observed across all the data to provide advanced contextual reasoning.
With increasingly large waves of data rolling in, IT can step up with some relatively easy solutions that have immediate and obvious business value, highlighting the connections and similarities between specific people, places, and things and helping establish the most common and likely relationship patterns at-large. We will definitely be exploring this area in more depth.
(p..s. Saffron is really based on some heavy duty matrix manipulations, but I've been told the graph analogy holds up well enough).
For more on Mike's view on graph databases see