The goal of this project aims at establishing a deap institute (consortium) and building an AI-based system that is able to share interactive, instantaneous, and user relevant earth science…

Prairie View A&M University, a historically Black college and university (HBCU) and 1890 land-grant institution in Texas, received a $1.5M cooperative agreement from NASA's Shared Services Center awarded on March 6, 2023, with completion targeted for March 5, 2027. The project establishes a Deep Earth and Planetary Institute (DEAP) consortium and develops an artificial intelligence-based system to make NASA Earth science information more discoverable and accessible to broad audiences. The core research objective involves mapping natural language questions about water-related topics from non-scientists into database queries using semi-supervised neural machine translation (NMT), a machine learning approach that leverages both labeled and unlabeled data to improve translation accuracy between layman's terminology and technical database queries.

The contract deliverables include a Docker image containing the semi-supervised NMT-based question-query mapping system and an integrated online website demonstrating the solution through NASA's Earth Information System for stakeholder engagement. No set-aside designation applies to this award. The project is designed to enhance research and educational capacity at multiple HBCUs and minority-serving institutions, promote underrepresented minority participation in advanced research, strengthen partnership with NASA, and build workforce diversity for the nation's science and technology sectors.

The goal of this project aims at establishing a deap institute (consortium) and building an AI-based system that is able to share interactive, instantaneous, and user relevant earth science information to allow NASA science to be more discoverable and more accessible to a broad audience. The research objective of this project is to map water related questions from non-scientists into database queries. IT can be defined as a sequence-to-sequence mapping problem, where the source sequence includes layman's terms on the water related questions while the target sequence will be most relevant database queries. We plan to complete this task by developing a novel mapping method based on low-resource neural machine translation (NMT). NMT has emerged to significantly promote machine translation with end-to-end models that automatically translate a source language to a target language through mapping source sentences to target sentences. Because large amount of data like pairs of annotated question and queries may not be available, low-resource NMT techniques will be developed for this project. Specifically, we plan to build a semi-supervised NMT based on our preliminary works on semi-supervised learning and natural language processing. IT will enhance the mapping performance by leveraging large amounts of unlabeled data. The deliverables include 1) docker image of semi-supervised NMT based question-query mapping 2) online website that is integrated into the ongoing earth information system that allows US to demonstrate the proposed solution to stakeholders at NASA and the community. The proposed project will greatly improve the research and educational capacity at multiple hbcu/msis, enable broad participation of underrepresented minorities in advanced research, build strong partnership with NASA, and provide the much-needed diversity in the next-generation workforce for the nation.