IBMers in Action

Watson Goes Global: How IBM Research – India is Advancing Cognitive Computing

Ramesh Gopinath, Dir., IBM Research – India

Ramesh Gopinath, Dir., IBM Research – India

A lot of the development work on IBM’s Watson, the computer that defeated two former grand-champions on the TV quiz show Jeopardy, was done by a small team of scientists and engineers in New York, near IBM Research headquarters. But you may be surprised to learn that some of the essential components of IBM’s first commercial product based on the Watson technology came from IBM Research –India. As director of the India lab, I’m very proud of that achievement.

The contributions from scientists in India demonstrate the value of having a global network of research laboratories. IBM has 12 labs in 10 countries, with a total of about 3000+ professional researchers. Today (Oct. 8 in India), when we convene a Cognitive Systems Colloquium  at the IBM Research –India offices in Bangalore and New Delhi, it will show once again the benefits of having a tightly integrated global research organization. The colloquium, one of eight such gatherings being conducted this year at IBM’s global labs, is designed to support the company’s comprehensive vision of the future of computing, which we call the cognitive era. Watson is the first step into that era.

To learn more about the new era of computing, download a free chapter of the book, Smart Machines: IBM’s Watson and the Era of Cognitive Systems, by IBM Research Director John Kelly, at http://cup.columbia.edu/static/cognitive.

SP-Watson2The Cognitive Systems Colloquium  in IBM Research – India will explore the fascinating realm of  “Biologically Inspired Cognitive Systems.” The colloquium will bring together leading academicians and thought leaders from the industry and government. Their domains include natural language processing (Prof. Pushpak Bhattacharyya – IIT Bangalore), perception engineering (Prof. Sumantra Dutta Roy – IIT Delhi), social listening (Prof. Vasudeva Varma, – IIT Hyderabad), Lovasz function (Prof. Chiranjib Bhattacharyya – IIS Bangalore), and spatio-temporal abstraction (Prof. Balaraman Ravindran – IIT Madras).

I myself am an example of the global and mobile nature of the tech industry’s talent pool. After growing up in India and receiving my undergraduate degree at the Indian Institute of Technology, Chennai, I got my PhD. at Rice University in Texas, and then worked for IBM Research in New York for nearly 20 years before becoming Director of the India Research operations late last year.

I arrived here in India after the Watson work had already commenced, but knew the whole story because of my involvement in it from the New York end at first. Since the lab was established with an office in Delhi in 1998, one of the focuses has been on developing technologies for IT service delivery—the many contact centers and business-process service centers that are located in countries all around the world.

We had scientists working on machine learning, natural language processing and text analytics–developing innovative solutions that helped IBM run its service delivery centers in India and elsewhere around the world. One such solution was a project called Voice of Customer Analytics (VOCA), a technology for deriving valuable business insights from textual portions of customer satisfaction surveys, call logs and other artifacts of service delivery interactions.

After Watson won on Jeopardy, people from the India lab got in touch with their counterparts in New York to adapt Watson technology to contact centers. At around the same time, back in New York, as part of a services innovation team study, I presented a new research initiative aimed at applying Watson-like technologies to services . This proposal was well-received by the executives and the assignment was taken up by the India lab. They commenced work on a project, code-named Holmes, for Sherlock Holmes, Dr. Watson’s partner-in-crime-fighting. To help get the India research team up to speed, our colleague, Eric Brown, who heads the Watson research project, invited several of them to New York to learn about the technologies by intermingling with his scientists.

The Holmes project started as a collaboration between teams from IBM’s Global Process Services business unit and Research with the goal of creating systems based on Watson technology for helping contact center agents respond to customer service queries faster and better. This necessitated the crafting of a different notion of questions and answers than the one employed on Jeopardy. Instead of posing answers and expecting responses in the form of a question, the Holmes technology had to be able to answer more open-ended questions. Example: “The battery of my iPhone is draining too fast. What do I do?”

The Holmes effort soon expanded in scope and size. It fed into work being done by the IBM Watson Solutions group, which was busy developing the first commercial product based on Watson. The product was released to the market in May 2013 as Watson Engagement Advisor. Enterprises would now be able to engage their customers better, faster, anywhere and anytime using automated self-assist solutions.

Technologies developed by India research lab were central to the new product. The India team enhanced the Watson technology for finding resolutions to complex customer issues, where the ‘answers’ can be steps, procedures, paragraphs or snippets of text.  The agent-assist technology enabled contact center agents to quickly find answers to complex customer questions, e.g. for technical and product support issues. With automated self-assist systems, this enabled end-customers to directly interact with Watson using natural language queries and find resolutions to their issues.

We are currently working on adding a conversational natural language dialog capability to Watson systems that will enable Watson Engagement Advisor solutions to engage customers more deeply and, in future, recommend the best possible products, offerings and service options to them based on other advanced analytics technologies from IBM Research. Natural, intuitive interfaces like conversational dialog will be a crucial pillar of cognitive computing systems.

But that’s just the beginning of our work in cognitive systems. I expect the team here to play a major role in advancing Watson and Watson-like technologies in the months and years ahead. This is an opportunity for our scientists in India to make a mark globally, not just for IBM but for all of society.
____________________

This post originally appeared on A Smarter Planet blog, and was republished for IBM Jobs Blog. Apply now. Be an IBMer.