Review from the Clouds

Inspiration

Recently, I bought a new phone. When trying to buy accessories for my phone, I was bombarded with multiple choices online and wondered if there was an easier way to see how other consumers felt about the products, other than the 5-star rating that many online platforms have picked up.

What it does

Our product extracts reviews and uses various Natural Language Processing techniques to identify keywords that many consumers associated the product with. Then, it converts this data into a visually appealing and easy to understand Word Cloud.

How we built it

We wrote all of the code in Python. We incorporated various libraries including Pandas, nltk, and keybert to help us work with the large amounts of data. For the User Interface, we used tkinter to create a simple app that creates pop ups for users to access the product.

Challenges we ran into

We struggled a lot to find the right libraries and NLP techniques that suited our needs. We went through many lists online (and installed a bunch of libraries) to get to the combination of libraries we ended up using. We also struggled to extract information from the datasets that we chose. We weren't very experienced with pandas (we didn't know of it until a few hours into the project) so we tried things like moving the data into a Word Doc and extracting it from there.

Accomplishments that we're proud of

This was our first time working with Artificial Intelligence so we are proud that we were able to get a working product even though it was relatively simple.

What we learned

One thing we struggled to figure out in the beginning but got towards the end was installing libraries. We got the chance to use the Command Prompt to install required libraries which ended up being much smoother and simpler than finding the right website online.

What's next for Review from the Clouds

One thing that may be a good step from our current product is making the application a Google Chrome Extension. That way, users could look at the Word Cloud Reviews of many products side by side at the same time to compare how other users thought of the product in question.

Built With

• kaggle
• keybert
• languagetool
• nltk
• numpy
• pandas
• pil
• python
• summa
• tkinter
• wordcloud