4 ms·
You're totally right about the pains with ingredients normalization - I've spent at least two months making scripts to scrape recipes, and it will only match th
by papa_bear 13y ago
You're totally right about the pains with ingredients normalization - I've spent at least two months making scripts to scrape recipes, and it will only match the ingredients accurately about 80% of the time, leaving a lot of work in hand curating everything it scrapes. I ended up trying to promote users to enter recipes by offering a month free every time they entered 5 good recipes. I built up a decent recipe database this way, but it does need to be much larger to better target people's tastes.
And I'd love to do the grocery store monetization idea, but trying to form partnerships with the grocery chains always seemed like a daunting time commitment. Maybe I'll send out some feelers this week.
- bernz 13y agoYeah, my ratio was about 80% and given the volume of recipes I wanted, that wasn't good enough. So I started coding an normalization solution that would highlight questionable recipes and make it easy to reformat them with structured data, but even that was really hard. It was honestly around that time that I lost interest and ditched the project.
- stmchn 13y agoHave you considered using Open Recipes? https://github.com/fictivekin/openrecipes https://github.com/fictivekin/openrecipes I saw it on HN a couple weeks ago and it's basically one big project to scrape and index food blog recipes. It doesn't include preparation instructions but it uses schema.org's recipe format, which includes ingredients and other useful things: http://schema.org/Recipe http://schema.org/Recipe