IBM Watson APIs hold key to broader cognitive computing use
One of IBM’s chief goals in 2015 was to make more of a market for its Watson cognitive computing system. While new machine learning technologies are spurring a resurgence of artificial intelligence (AI) capabilities in general, Watson seems somewhat alone as a cognitive champion. It still faces hurdles, but the company made some advances on its cognitive strategy during the course of the year — and an expanded set of IBM Watson APIs were central to that process.
The year has seen a parade of new Watson undertakings, ranging from early plans to connect Watson healthcare systems to iPhone users to last week’s announcement of a Watson-driven online shopping and product recommendation engine by outdoor clothing maker The North Face. But further progress in selling Watson to the corporate masses rides, in part, on IBM’s efforts to simplify the task of programming the system, in order to help take it out of the ranks of supercomputers and turn it into a more accessible platform.
Those efforts have centered on moving Watson to the cloud and connecting it to IBM’s Bluemix platform as a service (PaaS) environment for development. Turning Watson into an a la carte menu of accessible cloud services in effect reduces the large set of Watson application capabilities into bite-size portions for both customers and business partners.
Watson has been used in applications ranging from medical purposes, such as improving cancer diagnoses, to automating customer service processes to various kinds of research initiatives, according to Dave Schubmehl , an analyst at IT market research company IDC. Still, he continued, people need an easy entry to Watson.
“People don’t necessarily want to buy a million-dollar system to run Watson,” Schubmehl said. “But PaaS is well-suited for cognitive platforms. People can use Bluemix services and start working with one or two APIs, rather than use the whole system.”
He added that IBM’s March 2015 purchase of AlchemyAPI — a deep-learning AI technology startup — was also notable, as it brought to Big Blue a popular set of developer APIs that can help Watson in areas beyond machine learning applications. “AlchemyAPI had developed a business running on Web services. It really opened the door for IBM to start thinking about offering the rest of its services in the same way,” Schubmehl said. The acquisition also has vastly increased the ranks of developers available to work with Watson, he noted.
Profiles in data via Watson analytics
Using the IBM Watson APIs, applications can be created in pretty quick order, according to Michael Hussey, CEO and founder of StatSocial Inc., an Internet services firm based in New York that provides social media data to companies looking to analytics on the demographics of followers and other people in their “social audience.”
Among other traits, Watson can take free-form text and analyze keyword phrases to gauge people’s sentiments — useful information to add to a customer profile, Hussey said. StatSocial maintains its own database of text from blogs, tweets and other social outlets. It passes that material to the Watson Personality Insights service in the Bluemix cloud via an embedded API, “then we get back scores on profile types for consumers,” he said.
Hussey said things were up and running in less than two months. He originally saw Watson cognitive computing as ”a concept or marketing vehicle” — but as he has gained familiarity with the system, he has also gained a better understanding of its applicability in analytics applications.
“I had no idea a year ago how this would apply to what we’re doing,” Hussey said. “But the notion of cognitive computing is similar to what we do with people’s social footprints. We organize unstructured data — their social footprints — into a structured taxonomy. ” In the same way, Watson applications — in one way or another — can bring structure to unstructured data, he added. “For us, it’s a whole new level of texture for the data.”
Taking the time to train Watson
The task of programming Watson wasn’t intimidating to Rainer Baumann, head of shared information services at reinsurer Swiss Re, which is based in Zurich, Switzerland. But in an email interview, Baumann said that what does take dedicated work is teaching the system. During that phase, the system is repeatedly fed information, and analytical algorithms and models are adjusted in an effort to achieve the most accurate results possible in terms of predicting insurance trends and risks.
“We need to train the machine for a specific domain use case by preparing a large data set that is applicable to that domain,” Baumann said. What then follows is ongoing interaction between Swiss Re data specialists and Watson. That work is needed in order for the system to learn and understand specific questions, and provide accurate, evidence-based answers, he said.
Such efforts are just fine with IBM, which is continually looking for customers that can help the company train Watson for applications in new industries. Overall, IBM has estimated that more than 77,000 developers have worked with its Watson Developer Cloud platform to pilot, test or deploy new business applications. Much of its effort in the year to come will focus on moving that effort into hyperdrive and further ramping up use of the IBM Watson APIs to help drive more sales of the system.
This entry passed through the Full-Text RSS service – if this is your content and you’re reading it on someone else’s site, please read the FAQ at fivefilters.org/content-only/faq.php#publishers.
SearchBusinessAnalytics: BI, CPM and analytics news, tips and resources