Ever had the trouble of deciding what to get at a restaurant? Dishes scrapes Yelp's reviews, extracts references to dishes, and performs sentiment analysis to help you determine what's best at a restaurant. Currrently, only popular SF restaurants are supported.
- More robust dish name matching from reviews.
- Disregard adjectives/adverbs in dish names.
- Add detection for references with just one character off, such as apostrophes, plurals, spelling errors, etc.
- Build a new sentiment analyzer model based on scraped review data, with the validation being rating score.
- Improve README and write article to describe how dishes are ranked, with a little bit on sentiment analysis.
- Improve recommendation algorithm.
- Update rank order algorithm.
- Add support for more cities (NYC, LA, Vancouver, Chicago)
- Introduce comparisons between similar dishes across restaurants.