July 14, 2015 | Written by: IBM Research Editorial Staff
Share this post:
Worldwide, approximately 800 million people suffer from malnutrition. That’s about one in every nine people living on earth. How can we significantly affect this statistic given limited resources and an unequal distribution of farmlands and wealth around the globe?
One significant approach to reducing that number requires a major increase in yield production. Optimizing fertilization can increase yield by up to 50 percent, but it requires knowledge and resources that local farmers might not have access to. Our Fertilyzer app, one of the winners of the recent IBM Watson + GBS Cognitive Challenge
, just might hold the answer to helping farmers increase their yields dramatically.
An internal IBM effort, the Challenge galvanized employees to develop Watson-powered cognitive apps using Bluemix, IBM’s cloud platform for building and managing mobile applications. Our work on Fertilyzer was inspired by ongoing talks we have had with one of our agricultural services clients.
Fertilyzer was awarded “Best Mobile Experience” by Challenge judges for its intuitive interface, simple access to multiple information sources, and easy-to-understand data visualizations. But this Android app is more than just great to look at. Put together by an Israeli team of IBMers led by Roi Zahut from IBM’s Global Business Services
, fellow GBS employees Matan Mashiah and Yosef Ben-David, and Matan Ninio and myself from Research
, Fertilyzer is a cognitive mobile app built using Bluemix
. It taps into Watson technology to recommend optimized fertilization plans and track their progress.
The Fertilyzer model allows farmers to consider weather conditions, soil types, and field history to generate fertilization plans and predict cost, yield, and profit. Once all the predictions and related plans of action are ready, Fertilyzer then uses IBM Watson Tradeoff Analytics
to help the farmer make the right decision.
The app also includes a dashboard that helps keep the farmer updated on agricultural news and plan progression. Fertilyzer reads through an immense range of news articles and tweets and sends relevant news and alerts to the farmer when it finds something important, like a forecasted price drop or an upcoming storm, so farmers can act on new information.
Let’s say a farmer is planning an upcoming season for a specific crop, such as a russet potato variety. Fertilyzer’s connection with Watson Tradeoff Analytics can help him weigh different options, while stressing certain variables, such as profit, and filtering out others, like risk. But if news starts to trend before planting about an expected potato price crash in his region, Watson’s news filter can send an alert to the farmer’s Fertilyzer app, thereby allowing him to adjust crop plans before it’s too late. Ultimately, more-efficiently grown crops will improve yield, providing more food to more people. In a small but substantial way, we hope Fertilyzer will help communities tackle hunger and malnutrition.
Our next steps for development are to test the tool and gather more data. We’ve built a great prediction model that takes into account a wide variety of environmental parameters, but now we need to see how it will work in the field (pun intended). We also plan to extend the tool’s news and social media services to support more languages and learn how to detect even more events.
For more information about Fertilyzer, contact me.