product reviews analysis

Inspiration

help sellers to save time going through reviews and identify key customer insights on their products

What it does

Read customer reviews on a product and identify key topics that the customers talk about and the sentiment

How I built it

I collected Amazon product reviews data, clean the data, build the topic model and identify the top words in a topic and the sentiment to show users

Challenges I ran into

1. Cleaning the data
2. Identify irrelevant topic words and manually remove it

Accomplishments that I'm proud of

1. I figured out that using a tri-gram of text features work best to provide meaningful topic words

What I learned

1. For NLP, it's good to manually analyze the model prediction to see if I can manually adjust the data to help the model learn

What's next for product reviews analysis

Built With

• natural-language-processing
• sentiment
• topic-modelling