Internet protocol design discussions often take place in the Internet's standardization body, the IETF. Many of these discussions happen via emails, and discussions at meetings are transcribed. The IETF's email and meeting minute archives therefore constitute a large body of very useful text, containing valuable information about the decisions that were made.
Natural Language Processing (NLP) using Machine Learning could probably be used to automatically distill design decisions from IETF texts. This could, for example, make it possible to present a protocol header along with information about why header fields look the way they do.
The goal of this thesis is to create a starting point for such automatic text parsing, by manually trying to understand some protocol design decisions from this text archive, and collecting relevant information about this reading-and-understanding process.