Wimbledon nets analytic insight from Watson cognitive technology
In a previous blog post we looked at how IBM is building the “Wimbledon Digital Assistant” based on Watson technology, which has been trained to answer questions about the Wimbledon championships. This conclusion describes how we dealt with two types of data that Watson needed to understand—general knowledge and complex statistical data—while providing a single user experience.
To meet these differing needs, we built a Watson solution with two halves; one half focuses on general knowledge questions, and the other half concentrates more on statistics and numbers. We like to think of this as Watson's “right” and “left” brains. A common user interface across these two halves means the user of the system just types a question. And just as our brains automatically process the different characteristics of its two cerebral hemispheres, Watson seamlessly handles the interaction of its right and left brains without the user being aware.
Watson is a hypothesis-based system, meaning it first forms a set of hypotheses about what the possible answers to a question might be, and then determines its confidence in each answer by examining the supporting evidence for that answer. For Wimbledon, our Watson system passes the user's question to both the right-brained "statistical" area and the "left-brained" language region, and uses the answer confidence levels to determine what is given back to the user.
This same technology and basic approach can be used in many areas of general business. For example, banks and insurance companies are beginning to implement Watson systems to help answer customer questions about their complex products and services. Watson makes it possible to have an “Ask Watson” dialogue box on websites for customers to make inquiries and resolve problems on their own, which is often much more convenient than phoning a call center. The medical field is also using “Watson Oncology Advisor” to help diagnose and treat cancer.
As demonstrated in both Jeopardy! and Wimbledon, Watson can be trained in the ways language is used to convey different meanings. For example, in tennis the word “love” usually doesn’t mean an overabundance of affection between tennis players, but instead a score of zero. And in banking, Watson needs to understand that a British “current account” equals the American "checking account." We can train Watson to understand the unique way in which language is used in different domains so that it acts naturally when asked a question.
Looking to the future, we can envision a variety of opportunities to exploit cognitive technology. Watson’s capabilities now span areas as diverse as recognizing objects and faces in photographs, translating a speech recording into text, detecting the personality type of the author of a document and scouring the web for sources of news.
The way these other sources of Watson cognitive technology might be applied in both future Wimbledon championships and business at large is limited only by our imagination. Most exciting of all, these cognitive services are available for anyone to tap into on the Watson Developer Cloud—so building a "Powered by Watson" app is very easy.
So what can you do with Watson? How will you become inspired from our Watson work at Wimbledon to do something similar for your business? To find out, you or a developer friend can register on IBM Bluemix and create your own "Powered by Watson" app using the free cognitive services in the Watson Developer Zone.