It would need to be automatically collated by an AI with access to the Internet Archive that is probably quite a bit more sophisticated than what we currently have. DMOZ is the closest curated thing I can think of.