Topic modelling using Eccommerce Company reviews

PROJECT OVERVIEW

Goal: Create a review topic app that allows users (customers or Company affiliate) to drill down reviews based on the review sentiments or priority score

Companies reviews Used: The three ecommerce company reviews used are Amazon,Zalando and Outfittery

Data Source: Reviews were scrapped from TrustPilot

KEY MILESTONE

• Scraped reviews from Trustpilot using Beautifulsoup
• Stored clean data in AWS s3 , read it into AWS comprehend for topic modeling and sentiment analysis and stored it back to S3
• Discovered other python library for obvious spelling correction and expansion of English contractions e.g don’t == do not etc

ARCHITECTURE

DATA: BEFORE & AFTER Screenshot

DEMO VIDEO

TOOLS USED

• Key Python Libraries ( Pandas, Numpy, Matplotlib, re, Streamlit)
• Other Python Libraries (Contractions, SpellCheckers)
• AWS services (AWS S3, AWS Comprehend)
• VS Code
• Jupyter Notebook
• UI APP powered by Streamlit