Ghazaleh Kazeminejad

redshred

Modeling and Simulation, content and software developers are cutting and pasting work instructions from manuals , despite the availability of tools from Google Cloud, Microsoft Azure, and AWS. COTS tools work great for typical language but have limiting factors when it comes to tech data. First, the COTS tools require users to have NLP or coding know-how. Second, is that the complex tech data in Technical Orders (TO), Technical Manuals (TMs), DoD forms and other documents create trouble for these COTS tools as they highly use domain specific formatting, language, and jargon. Third, the tech data have complex infographics which aren't handled or treated as blind spots. RedShred treats documents as a database, effectively turning data trapped inside flat documents into usable structured information. RedShred approaches this process through document understanding by visually, textually, and semantically analyzing documents; much like a person when reading or skimming a page. This capability provides an AI-powered database that accelerates the development of knowledge-powered applications like AR/VR/MR, machine learning, and smart assistants. While ready for any domain, RedShred is also specialized in maintenance language that confounds off-the-shelf commercial NLP tools. RedShred is differentiated by SBIR-derived and patented technologies for extraction, linking and document analysis. RedShred has received 9 Phase I and II SBIR awards from USAF, NIST and NSF and commercialized the RedShred Platform through these projects. RedShred has solutions that are adapted to the military domain and provides a comprehensive documents-as-a-database platform that lets the developers integrate models and capabilities from RedShred document understanding tools, open source, and COTS products.