Automated Understanding: Digital Reasoning Cracks the Information Maze
March 4, 2011
I learned from one reader that the presentation by Tim Estes, the founder of Digital Reasoning, caused some positive buzz at a recent conference on the west coast. According to my source, this was a US government sponsored event focused on where content processing was going. The surprise was that as other presenters talked about the future, a company called Digital Reasoning displayed a next generation system. Keep in mind that i2 Ltd. is a solid analyst’s tool with technology roots that stretch back 15 years. (I did some work for the founder of i2 a few years ago and have a great appreciation for the case value of the system for law enforcement.) Palantir has some useful visualization tools, but the company continues to attract attention from litigation and brushes with outfits with some interesting sales practices. Beyond Search covered this story here and here.
ArnoldIT.com sees Digital Reasoning’s Synthesys as solving difficult information puzzles quickly and efficiently because it eliminates most of the false path or trial-and-error of traditional systems. Solving the information maze of real world flows is now possible in our view.
The shift was from semi-useful predictive numerical recipes and overlays or augmented outputs to something quite new and different. The Digital Reasoning presentation focused on real data and what the company called “automated understanding.”
For a few bucks last year, one of my colleagues and I got a look at the automated understanding approach of the Synthesys 3 platform. Tim Estes explained that real data poses major challenges to systems that lack an ability to process large flows, discern nuances, and apply what Mr. Estes described as “entity oriented analytics.”
Our take at ArnoldIT.com is that Digital Reasoning moves “beyond search” in a meaningful way. The key points we recall from our briefing was the a modular approach eliminates the need for a massive infrastructure build and the analytics reflect what is happening in a real time flow of unstructured information. My personal view is that historical research is best served by key word systems. The more advanced methods deliver actionable information and better decisions by focusing on the vast amounts of “now” data. A single Twitter message can be important. A meaningful analysis of a flow of Twitter messages moves insight to the next level.
Now we were not at this conference, but I had an opportunity to learn a bit about Digital Reasoning on my last road trip to Nashville, Tennessee, accompanied by one of our senior analysts, Dr. Tyra Oldham. Based on the information I picked up this morning, the talk at the conference on March 2, 2011, explained several of the points we picked up last year.
Here are three key points about Digital Reasoning’s approach:
First, real data is irregular like broken pavement, filled with potholes, and voluminous. As a result, a system must have methods for coping with “now” information that does not arrive in tidy, well-formed packages. Digital Reasoning has a content processing approach that solves most of the challenges of “real” data flowing into Syntheses. With transformation becoming a more costly line item in the information technology budget, the Digital Reasoning innovation is a big deal.
Second, key word search is pretty much useless at high volume flows of real data. Even a basic filter can yield too much information for an analyst to handle even with an arsenal of well=known tools. In our work, we don’t have time to read or craft queries to use the standard statistical tools we rely upon to make sense of health claims data, for example. The Digital Reasoning approach generates a summarization or “quick look” method that whittles big data down to what’s germane. The angle we seized upon in our review of the Synthesys system was the ability to focus on entities. We are usually interested in people, places, things, products, and events. Our world is filled with specifics. Digital Reasoning delivers specifics. This means we were able to rethink our approach to “big data” analytics. The “automated understanding” angle has attracted significant interest from companies I hold in high regard; for example, Fetch Technologies (yes, Googlers work at this outfit), Mitre (yes, I have done some work for this outfit when I lived in DC), and Cloudera.
Third, Digital Reasoning makes use of systems and methods that allow what we call “seamless scaling.” The Digital Reasoning approach allows an intelligence sy6stem to scale automatically.
How good is Digital Reasoning? During our for fee analysis, we were impressed. Listening to the information from our source, we have another source that sees Digital Reasoning as a company to watch in the exploding markets that require “automated understanding”, not pretty pictures, marketing hyperbole, and unmet expectations.
Check out Digital Reasoning at www.digitalreasoning.com.
Stephen E Arnold, March 4, 2011
Freebie but Dr. Oldham is convinced Digital Reasoning will buy us lunch the next time we are in Nashville.