Nmedia - Fotolia
Problems of big data analytics can overwhelm business managers and the businesses they run. Enter cognitive computing, which is an amalgam of natural language processing, analytics, machine learning and more . Growth in data from Web systems, intelligent devices and such is a major driver in the development of cognitive computing applications, according to technology consultant and author Judith Hurwitz.
Together with co-authors Marcia Kaufmann and Adrian Bowles, Hurwitz wrote Cognitive Computing and Big Data Analytics, a book published in March 2015 that makes a case for cognitive technology's potential, while at the same time acknowledging some challenges. SearchDataManagement spoke recently with Hurwitz, president and CEO of consultancy Hurwitz & Associates LLC, to gain perspective of the cognitive kind.
You point out in the book that elements of cognitive computing have been around for a long time -- some in the form of artificial intelligence. AI systems learned how to play games, and so on. But can we really build on that as the basis of a whole new computing category?
Judith Hurwitz: I think some of the early AI experiments were really designed to see, 'How far can we push things?' Questions were, 'Can a computer learn to play checkers?' and 'Can a computer learn to play chess?' And then when IBM got started with its Watson project, they started with the idea of a grand challenge. Could a computer understand the nuances and context of something like Jeopardy!? It's more than just a game. The reality is that something like Jeopardy! is actually a very complex environment for a computer. It includes categories, puns, slang -- so many different things that would not be obvious to a computer, and that require a lot of computation and analytics.
Jeopardy! was a challenge, and a proof of concept that sought to uncover whether you could train a computer on things where the relationships aren't obvious -- where the relationships require looking at patterns, looking at nuances and looking at context. That is where we started, I would say, with modern cognitive computing.
The body of knowledge a Jeopardy! contestant must master is vast, for sure. Mountains of big data can overwhelm the decision maker, too, especially in areas like medical diagnosis. That's a big drive behind the cognitive computing push, isn't it?
Hurwitz: Yes, the real challenge we're facing -- and this is the reason why cognitive computing is resonating -- is that we're in a world of more data, and more complex data, than ever. One of the early use cases for cognitive computing is around healthcare. It's just astronomical the number of pages of new research, new cases, new clinical trials and new treatments that the doctor must consider. New ideas are coming from across the world fast and furious.
Download this podcast -- Judith Hurwitz on cognitive computing and big data analytics
Cognitive computing involves natural language processing, data lakes, expert dialog modules and much more -- it's a lot of things. It seems as though there will be a learning curve for each element, ones with challenges that would need to be addressed with some patience.
Hurwitz: One observation I would make would be that it is early. If you look at each of the elements -- whether it's natural language processing, ontology, creating a corpus of data or ingesting data, applying analytics or machine learning, distributed computing, cloud -- each one of those has been approached as a topic in an area of endeavor. What you start to see is that cognitive computing is all of the above. When all of the elements come together at the right place, at the right time, with the right economic balance, then transformations and revolutions happen.
None of these things is magic. We're definitely at the pioneer stage. But I think what makes it different from a typical pioneer stage is that there is now a lot of expertise and knowledge in a lot of these different areas. It's not like AI 25 years ago where people sort of had a dream, but they didn't really know how they were going to be able to make it a reality. There's a lot of very advanced, complicated technology that is part of it. You have everybody from Facebook to Google to IBM to Hitachi to Cisco and Intel -- all these companies are investing in this.
Look at some first steps toward cognitive computing
Learn about virtual assistants that can aid in customer self-service
Listen to a podcast on IBM's Watson system