Beer Recommender System
A recommender system built from 10,000+ scraped beer reviews that uses similarity metrics and sentiment analysis to recommend beers based on taste profiles.
In this project, we scraped a beer review website for the top 150 beers ranked on their platform, collecting over 10,000 individual reviews. We then computed similarity and comparison metrics to understand the relationships between different beers based on their review profiles.
Using the similarity scores alongside sentiment analysis, we built a recommender system that suggests beers based on a user's taste preferences. The system can take a flavor profile — such as hoppy, malty, or citrusy — and recommend beers that best match those characteristics.
The central question driving this project was: what makes a beer awesome, and how can we use that knowledge to recommend great beers to new users? The combination of web scraping, NLP-driven sentiment analysis, and collaborative filtering techniques provided a comprehensive answer.