IBM: 8 Questions With the Company's New Big Data Honcho
One of the other important things, probably the most important thing around big data, is really a space where IBM leads by leaps and bounds and it’s what I call the data-in-motion space or streaming analytics. This is our InfoSphere Streams technology that allows the application of predictive analytics directly in the transaction path. It’s a combination of complex event processing and predictive analytics, which is transformative to things like next-generation telco mediation. You may have seen it first-hand in some of the examples we’ve talked about in health care over the years like Project Artemis or just in the fall Dr. Tim Buchman came out of Emory Hospital and talked about his work to transform patient monitoring and the intensive care unit’s use of InfoSphere Streams. But it’s relevant in connected car, in car-to-car and car-to-infrastructure connectivity. It’s transformative in oil exploration, helping our clients monitor yields off devices. And the applications will continue to expand as the Internet of things drives more and more volume and increases the importance of being able to take the signal out of the noise of all that the Internet of things will be generating. Can you go a little deeper with where cognitive computing comes into play and what we can expect to see in the future? When I talk about the advanced analytics space, we really possess a very robust range of analytics capabilities. We have things that are in what I would call the descriptive or business intelligence aspect, such as visualizations or what’s on the glass. Then you move into decision management, discovery and exploitation, then you move into predictive and prescriptive modeling–understanding what could happen, why is this happening, what action could I take.So when you put these things together you’re able to combine the speed and agility of putting predictive and prescriptive analytics in the business process, along with also being able to pull in the business record of everything our human journey has created as relevant information of learning and then being applied to a set of actions. The most compelling examples, and the ones I think are most viscerally understood, are the ones around health care. But that’s not all; there also is wealth management, simple consumer or citizen service, legal applications and also recipes. Watson can help you cook. It can create recipes and things that allow it to understand what’s being described in a recipe that is separate from the structured information. The sequence and the chemistry is defined in the language not necessarily the ingredients and measures. So it’s really when you need to make that jump to figure out what all have I learned and how do I apply that to what would be the best outcome of a situation. In those sorts of systems where the information is so varied, you have to train the system to understand the patterns and teach itself and create new algorithms from the information it possesses. I think one of the most relevant things that Watson does is it will give you a conversation of additional information it needs to give you a high confidence in its answer. So it will tell you, "based on what I see I know X, Y and Z, but if you answered these other questions I could give you a much more confident answer." I think that’s something that as productive as our current professions are, they’re also under duress. So having a bit of extra guidance and that kind of assistance is very relevant.
All of these things work on tabulating computing models and understanding the path statistics and being able to correlate into quintiles off of structured information. We do a superb job of helping our clients do this. We introduced new offerings like Predictive Maintenance and Quality to help people identify when they need to take prescriptive action as opposed to reactive action. These are all very strong fields, but there is a leap that you go through at some point to say "what is the record of mankind in this field and how do I correlate everything that I’ve learned to identify what’s best?" That’s where cognitive computing comes in. Because the record of man is typically in unstructured textual information; it’s not typically tabular. It does have tabular markers in it, but the nuance around the tabular information is important as is deriving context from that information.