Enterprise are enforced to comply to regulations by institutions such as governments. Large enterprise typically maintain individual policy documents. This arises the need to map the policies to different regulatory documents due to audits or for updates in either documents. Alyne is a startup that solves this problem with intermediate so-called control statements that maps to both: regulatory document paragraphs and enterprise policies paragraphs. The goal of this thesis is to investigate how far this mapping can be supported by the application of natural language processing and machine learning methods, in particular word embeddings.