Natural Language Processing Template Engine

In this presentation we discuss the completion of computational templates with parameters that are extracted from text specifications using a question answering system (QAS) [Wk1]. We outline the general method and then demonstrate it with several types of computational workflows: classification, latent semantic analysis (LSA), quantile regression (QR), random data generation (RDG) and recommendations. We show how to leverage the engine by bringing your own templates (BYOT). The demonstrated functionalities use the packages in the project ""NLP Template Engine"" [AAr1], which is based on FindTextualAnswer [WRI1, JL1]. We concentrate on data science (DS) and machine learning (ML) computations, but the described method and overall algorithm are applicable to computation workflows from other fields.