Continuing the IBM Big Data and Analytics event we come to Watson. Watson is designed to understand natural language, human-style communication. Watson then trawls through a potentially very large amount of material to create and score some hypotheses as answers for these questions and returns them. With each interaction it also learns what works, what the better answer is. IBM has been commercializing Watson through their transformation plays where they work with large numbers of companies to try and solve big problems, by selling it to their enterprise customers and through the Watson Ecosystem where they have some thousands of use cases submitted and under development.
Watson’s ecosystem includes the Watson Developer Cloud, the Watson Content Store where IBM is adding new content sources to the pool and the Watson Talent Hub with subject matter experts. The Watson Content Store is being driven both by IBM and by requests from partners. The content is focused in different areas like travel, healthcare, retail etc. Each content provider can be sampled by developers before they pay for it – while IBM is managing the relationships with these third parties, developers have to pay for some of the content.
A good use case for Watson has a few specific characteristics:
- Obviously it has to have a business case and a need for question and answer style interaction.
- It needs Watson now rather than needing business rules/analytics/decision management now and Watson (perhaps) later.
- You already know (or at least have some idea) what content will be used to answer the questions and this content is available.
IBM is very focused on developing a self-service, partner-friendly process for signing up for Watson, loading up content, testing and development. IBM also acts as a minority investment partner in some of these Watson developers to help move these systems along.
As an example, Welltok talked about their adoption of Watson in their wellness management platform. They really wanted the kind of cognitive computing technology Watson offered but they also wanted a robust Natural Language technology platform from an established company like IBM.
The product as developed drives personalized interaction with consumers to provide wellness advice. The product was developed mobile-first as this was the primary use case and this impacted the kind of questions people asked. The team also needs to support the continuous training of Watson on their unique content plus it needs to continually update the content being analyzed. Besides the Watson API they also use predictive analytics as part of the response and enrich the responses from Watson to provide more prescriptive answers.