Machine Learning (ML) has offered innovative benefits in automated content analysis and discovery.  In the commercial space, sophisticated Deep Learning (DL) neural models have enabled systems capable of better understanding queries and the content being searched.   More recently, an AI system was able to comb through a large dataset of scientist literature to discover a new novel antibiotic.  While it might seem that understanding our content is a solved problem, these breakthroughs have come at the cost of the large amount of data required for training the neural networks so that they correctly process technical terms and specialized language use.  It is often the case for Military applications, that the amount of data available is far more limited.  The question becomes: How can we leverage recent advances with sparse data to train the warfighter?

This paper presents a system utilizing the key advances in textual Deep Neural Embeddings by leveraging transfer learning from a larger corpus in order to automatically understand content topics.  This automation of understanding allows for enhanced automated meta-data to be annotated with fine granularity without increased book keeping for content developers. This understanding allows for content to be automatically indexed, annotated and modularized, aiding in training content reuse and adaptation without additional task load on instructional content designers.   Individual varied instruction is envisioned in the Department of Defense's future warfighter training systems. Creating and using fine-grained meta-data for instructional content is a necessary enhancement to support individual varied instruction because it helps to answer how content can be modularized, what learners can be expected to know after using the content, and which content should be presented to optimally teach and train learners in different contexts, backgrounds, and performance.