Academic project, written for course "Natural Language Processing".
This was a Kaggle-like competition between students. We were given a test data-set of annotated product reviews. Each review mentioned some feature of a product: price, quality, etc.
The goal was to produce a solution, which automatically annotate previously unknown reviews: whether a given review is positive, neutral or negative with respect to price, quality, etc. The output looked like this:
price: neutral
quality: positive
personnel: negative.
-Implemented parser from given json format
-Designed a solution with use of NLTK and scikit-learn