Information Extraction for New Emerging Noisy User-generated Micro-Text

This proposal presents an opportunity to address many challenges of named entity recognition (NER) and related information extraction techniques when applied to noisy user-generated microtext, or “chat” data, in support of United States Air Force and Air National Guard intelligence analysts leveraging such sources in their missions. Given the breadth of chat data platforms and dialects coupled with the need to address low resource settings, we propose a three-pronged feasibility investigation: one focusing on deep networks with end-to-end semi-supervised learning, one applying higher-order deep learning techniques, and one using a hybrid approach of the two. Evaluated approaches will consider the typical metrics of precision, recall and F-measure, as well as runtime performance and operational metrics including the effort required to develop and maintain training sets. Software will be developed using agile methods, evaluated, and if found to be feasible, ultimately integrated into active operational transition paths.