Abstract: User-populated forms may contain valuable information for use in decision making. Large, multi-page or multi-field forms often contain the greatest amount of valuable information, but can suffer from discouraged users that may not accurately populate or submit forms due to the total time required, reducing data capture and threatening data quality. At the NASA Jet Propulsion Laboratory (JPL), form field recommendations are provided using machine learning to spacecraft engineers within internal reporting tools, releasing valuable time previously spent on repetitive administrative tasks, improving organizational efficiency. These form field recommendations are provided by Henosis, a cloud-native, lightweight Python-based recommender framework developed and used internally at NASA JPL. Henosis, together with scikit-learn, Elasticsearch, and Amazon Web Services (AWS) S3, brings together model training and testing, storage and deployment, and querying under a single, easy to use framework. Henosis provides Data Scientists with a straight-forward and generalizable environment in which to train, test, store, and deploy machine learning models for form field recommendations, while also providing software engineers with access to recommendations via a REST API that can be easily queried and integrated across different enterprise applications. In this talk, we describe the motivations and concept behind the development of Henosis and introduce how it can be used to quickly provide form field recommendations in large multi-page or multi-field forms.
Bio: Valentino Constantinou is a data scientist at NASA's Jet Propulsion Laboratory (JPL) with the Technology User Evaluation and Infusion Office, where he develops recommender systems and open source software. Prior to joining JPL, he graduated from Northwestern University's Master of Science in Analytics program in 2016 and from the University of Tennessee with a Bachelor of Science in Economics in 2014. In addition to his work with the recommender system, he is working on tools for the Foundry mission formulation office and the Microwave Limb Sounder team using natural language processing and graphs