Feature extraction for sentiment analysis

Question

I am working on a group project for my capstone course and we have been tasked with creating a sentiment analysis tool with Python business logic and (L/W)AMP everything else.

We have good feedback for every part of our project plan except for feature extraction. One of our advisors insists that we should have ~15 different kinds of features.

Currently we only use unigrams and are having a hard time finding others that are practical to implement with our small set of data (~50 items) and within our time limit (~2 weeks to fully implement).

What are feature extraction techniques that are useful for sentiment analysis and work on smaller datasets? They should be able to be implemented quickly or already exist in a Python library.

Answer 1

Have a look at these papers-

• Semantic Similarity - 1
• Semantic Similarity - 2

They came up with a good solution. Have a look maybe you will find something of your use in it. Plus try exploring these too

• vaderSentiment 0.5
  (VADER Sentiment Analysis. VADER (Valence Aware Dictionary and sEntiment Reasoner) is a lexicon and rule-based sentiment analysis tool that is specifically attuned to sentiments expressed in social media, and works well on texts from other domains.)
• Demo- Sentiment Analysis with Python
  (This is a demonstration of sentiment analysis using a NLTK 2.0.4 powered text classification process. It can tell you whether it thinks the text you enter below expresses positive sentiment, negative sentiment, or if it's neutral. Using hierarchical classification, neutrality is determined first, and sentiment polarity is determined second, but only if the text is not neutral.)

Hope it helps! :)