-
Home page, where users input Shopee product listing URL
-
Results page to show sentiments, emotions, and overall comments
Inspiration
When users visit a product listing, they are sometimes overwhelmed by the sheer amount of reviews available. This web-app resolves the issue by categorizing large datasets of reviews into bite-sized nuggets of information in terms of easily understandable data visualization of the overall reviews.
What it does
Sentiment Analysis
AI model analyzes each product review and classifies it into either positive/neutral/negative sentiments.
Review Summarizer
Summarizer model aggregates comments for each rating category and extracts the most frequent points mentioned in the product reviews.
Emotion Detector
Enables user to get an overall idea of emotions from the tone of product reviews, with colourful emojis symbolizing each emotion.
How we built it
NLP models (summarizer, emotion detector, sentiment analysis) was built using pre-made HuggingFace PyTorch models. Front-end was built using HTML, CSS, JS, and integrated with backend using Flask API.
Challenges we ran into
- Hard to debug integration between NLP model and front-end web application.
- Long time taken for model to evaluate emotions and summarize thousands of comments for the product reviews.
Accomplishments that we're proud of
Sentiment analysis, product reviews summarizer, emotion detector in reviews.
Log in or sign up for Devpost to join the conversation.