Coming Up For Air – The End of Sprint #1

Greetings from the finish line of Sprint #1! This summer, our teams within Extreme Blue internship program each have 12 weeks to build a new product from the ground up. The 12-week development period is broken in to 6 two-week sprints, the first of which we have just completed.

What a ride! Our project is called the Watson Document Ingestion Evaluation Tool (DIET). Team DIET consists of four students from around the country. Our names are Jack Cusick (Rensselaer Polytechnic Institute), Shannen Lambdin (University of Nebraska- Lincoln), Ben Lopez (The University of Texas at Austin, McCombs School of Business), and Anuj Sharma (Columbia University). After spending a couple of days becoming oriented with IBM and getting to know our teammates and mentors, we jumped straight into the deep end and got to work.

IBM US Extreme Blue Interns: Team DIET in their natural environment. From left: Shannen, Jack, Ben, and Anuj.

DIET is a Watson Oncology tool used to determine the “value” of an oncology medical journal article. Value in this sense refers to how well Watson ingests the material, and how useful that material is to our clients. This sort of tool is required as the corpus of medical content accessed by Watson now contains nearly a million medical journal articles. We accomplish this by using a large set of value metrics which are determined using natural language processing. Generally speaking, most of our value metrics relate to article recency and relevancy. Some metrics include: citation strength, author credibility, density of biomedical entities, and the types of semantic relations in the documents. The entire set of metrics is passed through a machine learning algorithm that uses several “ground truth” standards to perform feature selection and assign an importance weight to each metric.  We finally give each article a score based on a weighted sum of these value metrics.

The focus for the summer is on medical journal publishers. The ultimate goal is to create a domain independent ranking system that will aid IBM executives in making well-informed purchasing decisions during contract negotiations. This will minimize wasteful spending and the amount of little-or-no value content that goes into the Watson Oncology database. For content publishers that are facing increasingly fierce competition and price pressures, DIET will help make more informed decisions about what type of journals to keep providing to maximize their value.

Enough about DIET, let’s talk about Sprint #1! In addition to addressing the challenges of laying the foundation for a brand new product, the first sprint was also about learning how to work as a cohesive unit and become a team. This included learning a new task management system, exploring nearby restaurants, and making fun of one of our teammates’ Chipotle and cheese addiction. Looking back on the past two weeks, I’m not sure how our transition from ‘four interns that just met each other’ into Team DIET could have been any easier. Upon our arrival, it was clear that our Extreme Blue Lab Manager, Ross Grady, and our team of IBM mentors had done a ton of work to make our transition as smooth as possible. We’re certainly grateful for their continued help and support as we move forward on our development journey this summer.

With the first sprint complete, our team is looking forward to an exciting and (hopefully) restful weekend exploring the Triangle. You might find us exploring local breweries and coffee shops in Chapel Hill, sitting in centerfield at a Durham Bulls baseball game, taking in some culture at the North Carolina Museum of Art in Raleigh, or simply hitting the movie theater to see Jurassic World. Sprint #2 begins Monday, and it’s going to be a whole new ballgame- Wish us luck!

Thanks for reading!


Extreme Blue is IBM’s premier biz/tech innovation internship. Learn more at #ExtremeBlue

This article was first published on IBM Extreme Blue blog on Tumblr:

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.