web crawler / semantic classifier

(Working version). Project to refactor (TurboGears -> Django) webapp designed to create sentiment scores related to companies, products, movies, etc. Visit any number of URLs, follows links to any depth, scrapes text around user-provided keywords. The text is then analyzed using MIT's ConceptNet for conceptual proximity, then a simple classifier for positivity/negativity.