I’m working on a small side project in SWI-Prolog that scrapes and normalizes product listings for camping gear, and the boards I follow keep mixing in reviews of portable toilets. I came across this Kohree 5.8 Gallon Portable RV Toilet Review while figuring out what fields a listing should expose, and now I’m stuck on how deep to model the data. My plan is a small set of facts for capacity, dimensions and waste type, with DCG rules to pull those out of messy titles. The concern is that product attributes vary so much between listings that a rigid schema might break on edge cases, and I don’t want to over-engineer something that only needs to filter and sort. Should I lean on Prolog’s unification and backtracking for the fuzzy matching, or keep the parsing simple and push the messy parts into a separate validation pass? I’d appreciate hearing from anyone who has modeled product data in Prolog.