Most (if not all) indexes of measuring the scientific performance of researchers rely primarily on citations (such as impact factor, h-index). However, this is more a measure of how popular a certain field, paper or person is, rather than an explicit quantification of how the respective research piece has contributed.
The idea put forward is to 'create' (or adapt) an algorithm that searches through different research documents (e.g. papers in journals) and compare them with "standard' knowledge (e.g. renowned textbooks), and then conclude on how much each research piece has contributed to the scientific world (e.g. how much from each research paper ended up in a textbook).
A relatively easy implementation would be to use existing open-source plagiarism tools for the comparison of texts, and adapt it for our purposes.
P.S.: This project could have further ramifications, like textbook autocompletion rather than writing it from scratch.
Adapt a neural network and apply it to a QM book.
Natural Language Processing
Machine Learning
Enthusiasm
Human beings whose core principles are freedom, equity and logic.