5 tips for working with IBM Watson
Published January 30, 2015
LifeLearn, a Canadian company that focuses on knowledge mobilization for the animal health industry, recently built a mobile application powered by IBM Watson that serves as a diagnosis and reference tool for veterinarians.
Keith Winn, Vice President of Product Development at LifeLearn, spearheaded the project and offers his top five tips for organizations that plan to work with Watson:
- Don’t underestimate the effort required in content formatting and training. This is equivalent to the programming effort required in a traditional technology product.
- Train your first piece of content quickly, then revise your idea of how your application will interact with Watson. Rather than simple question/answer exchanges, consider more sophisticated strategies that include meta information, tagging, filtering, and behind-the-scenes processing. Your application probably will need to be more than Watson and your corpus.
- Consider combining and reformatting the content before you input the data into Watson. This will improve the value of Watson’s answers.
- Be deliberate, detail oriented, and relentless in your exchanges with IBM.
- Connect frequently with other Watson partners to exchange experiences and swap solutions.
Check here to read the entire case study and delve deeper into this project. Also, see a video of how Sofie, powered by Watson, works!