Enabled Intelligence proposes a Phase II program to research and trial various data conditioning and labeling processes and develop a recommended solution(s) for creating valid and reliable ground truth from NGA and open source data. The goal of the program is to provide NGA with deeper understanding of various labeling methods, processes, and tools and their impact on labeling efficiency and eventual AI model performance. NGA would be able to use this analysis to develop and implement future internal data curation and labeling strategies to build more reliable AI technologies to support various analyses and intelligence products. \n\n The developed data labeling processes could be used by NGA as a template to guide future data collection and annotation efforts. Additionally, during Phase II, Enabled Intelligence will create multiple labeled data sets using NGA provided source data in order to demonstrate the differences in the labeling processes across a number of quantitative metrics. By necessity, Enabled Intelligence will create multiple prototype AI models using the labeled data to test the sensitivity of the model performance to the different types of labeling approaches.
