While we see a decrease in accuracy in annotations returned from text when we include the synonyms generated by our rules, we do not attribute the decrease entirely to the synonyms themselves, as over half of the errors are due to interaction of synonyms and the stemmer utilized for dictionary...