This document summarizes a thesis on automating test routine creation through natural language processing. The author proposes using word embeddings and recommender systems to automatically generate test cases from requirements documents and link them together. The methodology involves representing text as word vectors, calculating similarity between requirements and test blocks, and applying association rule mining on test block sequences. An experiment on a space operations dataset showed the approach improved productivity in test creation and requirements tracing over manual methods. Future work could explore using deep learning models and collecting additional evaluation metrics from users.