Back to Projects
// Personal project
Sentiment Spark
I built a product review and feedback analysis tool that automatically classifies every submitted review as positive, negative, or neutral using a natural language processing model.
Tech Stack
PythonNatural Language ProcessingMachine Learning
How it works
Real reviews are messy, so I built the model to handle that directly.
- Built the classification model to handle real, informal review text, including typos, casual language, and mixed opinions.
- Avoided assuming clean, formal input, since real customer reviews rarely look like that.
Why I built it
Sentiment tools trained only on clean, formal text tend to break on real customer reviews, which are full of typos, slang, and mixed opinions in the same sentence.
How I approached it
I built and tuned the NLP model specifically against messy, real-world review text instead of clean sample data, so it holds up on actual customer feedback.
