An simple sentiment engine that identifies positive and negative sentiments and their associated entities. dependencies: -Tweepy -Couchdb -Couchdbkit -NLTK -PyML TODO: Implement views for all objects whose sentiments have not been extracted. Improve feature selection for SVM to reduce training time. Test hypothesis of using bounds like 'worst' and 'best' to create segmenting attributes based on distance in wordnet and a trained classifier. Generate actual product features by clustering. Speedup NLTK POS Tagging