semantic analysis with nltk
- begin with a pre-determined list of words/phrases, count the occurances of words nearby
- find clusters of words, which are likely to be found near eachother
- find the most common words, trace their occurance over various levels