As the next part of our blog series covering the Watson + GBS Challenge we’re sharing the development experiences of two of the five winning teams – part one here. So if you’re looking for detailed understanding and insights directly from the teams themselves, look no further. Enjoy, and feel free to reach out to the team leaders if you have any questions, or just add a comment below. Screen Shot 2015-06-17 at 2.44.07 PM Here’s a look at what Roi Zahut had to say about the development of Fertilyzer. What is Fertilyzer? Fertilyzer is a cognitive mobile app that uses IBM Watson technology to recommend optimized fertilization plans and track their progress. What Challenge does Fertilyzer address? Fertilyzer improves soil utilization and crop yield production in an effort to address malnutrition-a billion person global challenge. What cognitive API’s have you integrated into the app? We used the following APIs: Alchemy API – finding and classifying news and tweets IBM Text to Speech – dubbing news items Watson Tradeoff Analytics – helping the farmer choose the right fertilization plan Visualization Rendering (RAVE) SDK – visually present the performance of farmers around Which API do you think offers the most game-changing capability? In our opinion, the API that offers the most game-changing capability is the Alchemy text and visual API. It’s a robust service that allows anyone to perform advanced textual analytics on text in different languages, without the need for knowledge of Natural Language Processing (NLP) and Bayesian Networks. Fertilyzer uses Alchemy API to read through an immense range of news articles and tweets to find items that may be relevant to our users. This task alone would not have been possible without the textual analysis capabilities of this service. The task is complicated as each news item or tweet that goes into our system has to be related to agronomy or farming, a task that is hard to achieve without cognitive capabilities. Furthermore, we are using Alchemy API to look for specific target keywords (eg. “potato prices”) and check the sentiment that is associated with such terms. This allows us to issue real-time alerts to users in the geographical areas mentioned (also extracted by Alchemy) so users can act on new information proactively. How difficult would you say it was to build your application using the Bluemix platform? Building an application using Bluemix is relatively easy. We specifically went with two main back-end technologies: Java and NodeJS (over Node-Red boiler plate). Setting up an application was a piece-of-cake. The Node-Red boiler plate gave us an easy way to edit our code right in the browser and deploy it immediately (it takes up to 10 seconds to deploy your code). It provides a visual editor that allows you to build your application by describing the desired workflow. You can literally get from no app to a running web service in 3 minutes. Setting up a Java application instance was also relatively easy. All that is needed is to write your Java code as you normally would and use the CLI interface to push your changes to the Git repository. From there everything gets immediately deployed to your application instance on Bluemix. As a programmer from the pre-cloud era, I can’t stress enough how quick and easy it is to setup and modify code of an application – no more uploading files with FTP and restarting services. What advice would you give to a developer looking to build with Watson API’s for the first time? My best advice is to make sure you understand what the service is about, its advantages and limitations. From there, just go ahead and try to use it. Most of the services are very easy to use and are pretty straight forward. Watson services do not require you to read pages of documentation in order to effectively utilize them. Boilerplates, when available, are also very helpful. What do you see as the next step for your application’s development? The next step for Fertilyzer development is to conduct field testing and data gathering. We have built a great prediction model that takes into account environmental parameters such as predicted and historical weather, location, soil type, date and more to predict how different fertilization plans will play out. So it allows the user – a farmer – to choose a plan that gives he or she the best tradeoff between cost, profit, risk etc. Our model was built with knowledge and data from academic papers and best practices, and now needs to be tested in the field and against previously collected data. We are also planning to extend our news and social media services to support more languages and learn how to detect even more adverse events. For more information about Watson Fertilyzer please feel free to email Roi directly. Screen Shot 2015-06-17 at 2.43.51 PM Here’s a look at what Will Rosenfeld had to say about the development of Watson Patient Empowerment. What is Watson Patient Empowerment? Watson Patient Empowerment is a revolutionary app which enables both patients and health networks to be more involved in the process of selecting and vetting specialist care providers. What challenge does Watson Patient Empowerment address? One in three medical patients receive a referral to a specialist each year, however these patients are rarely afforded the opportunity to vet such caregivers based on the highly unique preferences he or she may have for an ideal provider. Watson Patient Empowerment puts the patient at the center of this process. Which cognitive API’s did you integrate?
 Watson Tradeoff Analytics is core to our app’s functionality and value proposition. In today’s world of option overload, users need a way to eliminate superfluous options and refine their choices in an intuitive manner. The Tradeoff Analytic’s Pareto algorithm and multidimensional GUI combine to accomplish both these feats. We also integrated the Alchemy API’s Entity Extraction service and the Question and Answer service to customize and answer common medical questions related to user searches. On a scale of 0-10 what was your overall team’s experience level using Watson services or Bluemix prior to this event (0 = none, 10 = expert user)? Zero. Our team came into this competition with only a vague understanding of the capabilities of the Watson services and the Bluemix platform. Armed with moderate experience building web apps with Ruby on Rails and SQL databases, we were able to get our site up and running in days, not weeks. 
 What advice would you give to a developer looking to build with Watson API’s for the first time? Dive right in. The APIs and SDKs available through the Watson Developer Cloud provide powerful tools which you can leverage today to create amazing experiences for your users, and strong value propositions for businesses. Documentation of the technical details is well developed and easy to follow. If you get stuck, help from experts is also available in the IBM Developer Community. Our group used the community to get our first service up and running, and we were able to complete our app with the documentation provided in the Watson Developer Cloud. For more information about Watson Patient Empowerment please feel free to email Will directly.

1 comment on"Watson + GBS Challenge, Chat with the Developers (2 of 3)"

  1. where are the links to the apps?

Join The Discussion

Your email address will not be published. Required fields are marked *