RestaurantReviewParaser

Extension of BetterFoodRecs

Given restaurant A, find the most positively reviewed items on the menu

Requirements

• OpenAI API Key
• Zembra API Key

Setup

Pre-requisites

pip3 install unidecode openai gdown gensim Sanic python-dotenv
gdown https://drive.google.com/uc?id=0B7XkCwpI5KDYNlNUTTlSS21pQmM
gzip -d GoogleNews-vectors-negative300.bin.gz

Create a .env file with:

REVIEW_COUNT=400 # number of reviews to analyze
PORT=8085 # port to run server on
OPENAI_API=API_KEY # openai api key
ZEMBRA_API=API_KEY # zembra api key

To start:

python3 app.py

Usage

Endpoint: /api/v1/fetch/
Body: {"location": "<location slug>"}

<location slug> = Google Maps location ID. Formats:
472717649119152494
https://www.google.com/maps?cid=472717649119152494
https://maps.google.com/maps?cid=472717649119152494
ChIJx0JMBTFV2YARbgnOgjJujwY
https://www.google.com/maps/place/The+Cheesecake+Factory/@32.76918,-117.1677887,17z/data=!3m1!4b1!4m5!3m4!1s0x0:0x68f6e3282ce096e!8m2!3d32.76918!4d-117.1656
0x68f6e3282ce096e

Cost

Zembra API Cost: $0.06 + $0.12/100 reviews

OpenAI API Cost: ~ $0.14/100 reviews

Total Cost/Request (400 review default): ~$1

Reducing costs: Instead of text-davinci-003, a lower-powered GPT model can be used to lower review parsing costs.