Abstract
Statement of Problem
Extracting technical requirements and finding the changes between latest and previous revisions of specifications is a very tedious and time-consuming task. Often taking valuable time and resources to gather what the differences or deltas are in a new specification. This project focuses on using Generative AI with RAG (Retrieval Augmented Generation) an Artificial Intelligence and Machine Learning technique to extract requirements from any technical specification and compare the differences between the latest version with a previous version of the same specification.
Sources of Data
Any technical specification with at least one previous revision can be used.
Conclusions Reached
RAG with Generative AI techniques cannot extract all the requirements from technical specifications as the inference output from the AI model is non-deterministic, therefore structured programming is better suited for the extraction portion. However, Generative AI with RAG can compare differences between two revisions of the technical specification after the requirements have been extracted. Using the combination of structured programming and Gen AI with RAG, we can extract the requirements and sort the differences between the specifications. Sections are also ordered from the most to the least number of differences.