Use NLTK English stopword list in initial implementation. Then make it possible to choose from other stopword lists from a database of stopword lists.