jamesliangg / uOttaHack-5-CensView

Finds the consensus of reviews of different products

uOttaHack-5-CensView

---

Finds the consensus of reviews of different products

Inspiration

Choosing which product to buy shouldn't be hard. After having bought multiple products online with "high" ratings only to be disappointed. We decide to make an app that helps shoppers make a more informed decisions.

What it does

CensView is a Chrome extension that condenses and summarizes customer reviews of a product page. Using Cohere's NLP, the app gathers the general consensus of the reviews to help customers make a more informed decision.

How we built it

The frontend is built using HTML, CSS, and JavaScript. The backend is coded in Python and Javascript, it receives the product URL, scraps the review data from the product site with Puppeteer, and uses Cohere's NLP to analyze the data. Communication between front and back is handled using ExpressJS and Flask.

Challenges we ran into

• Communicating between frontend and backend.
• GET and POST requests
• Finding selectors for Puppeteer

Accomplishments that we're proud of

• Created a working application
• Support for BestBuy and Amazon
• Teamwork

What we learned

• Creating chrome extensions
• Working with Cohere
• Full Stack Web Development

What's next for CensView

• Adding support for more storefronts
• Hosting the server for the extension online
• Improving the UI look

Running the Extension

You'll need two seperate terminal windows (terminal1 & terminal2)

1. Clone repository from GitHub
2. Run npm install in terminal1
3. Create environment
   • Mac - run python3 -m venv venv in terminal2
   • Windows - run py -3 -m venv venv in terminal2
4. Activate enivronment
   • Mac - run . venv/bin/activate in terminal2
   • Windows - run venv\Scripts\activate in terminal2
5. Run the following pip installs in terminal2

pip install Flask
pip install cohere

6. Create a file with your Cohere API token called dumby.py. The contents of the file should look like below:

token = 'API_TOKEN'

7. In terminal2 window run flask --app app run
8. In terminal1 window run node app
9. Can send a GET request to flask for array result, sample results below

Example GET Requests

• http://127.0.0.1:5000?siteurl=https://www.bestbuy.ca/en-ca/product/sonos-arc-sound-bar-black/14597172&website=bestbuy
• http://127.0.0.1:5000?siteurl=https://www.amazon.ca/dp/B09F1QQZM2&website=amazon