Algorithm for using NLP with extremely small text datasets

A quick analysis of current technological trends displays an increasing global tendency of turning towards data-centric solutions. The ongoing wave of wide spread implementation of machine learning algorithms and artificial intelligence techniques allows us to harness the power of data to derive solutions and innovations. However, one major constraint that these algorithms face is the requirement of large amounts of data to train the models over. This project attempts to solve that issue by implementing a custom algorithm that is capable of generating NLP models that return medium to high accuracy results despite being trained on a very limited number of text data points. The posited algorithm is tested against a custom dataset consisting of answers to a set of questions pertaining to fears faced in day to day life.