I discontinued it mainly because it’s unsustainable. The only reason it worked so well was that I crawled websites with a home botnet of laptops, then I had to manually merge this data, so I have a giant csv of all the beer names across the websites (then I had excruciatingly complicated logic to figure out how to “merge” same beers which are named differently). Then I had even more complicated logic to figure out which words from the bag of words is likely to be associated with an actual beer because a menu could be full of long descriptions or unrelated words, including one branch which detects if the menu is from a specific craft brewery so you can still get all the right beers even if you’re in an obscure craft brewery.
It did work better than literally any other beer app out there, and the accuracy was fine, but I noticed some merging bugs which I had to write increasingly complicated logic for (like how do you know Blue Moon from Miller-Coors is actually the same as Blue Moon from its subsidiary which Miller-Coors bought, and apply this to all other similar cases). The whole web scraping thing was also starting to get very tedious especially since I expected to make zero money from the app and my only real user is myself. Without the constant webscraping, new beers will not be recognized and my app would decline in quality, so I just kinda let it die. Maybe I will work on it again in the future.