This presentation will focus briefly on some current methods used for automatic Part of Speech tagging - statistical and rule-based. I will also give a brief overview of corpus linguistics.
In 1992, Eric Brill introduced a hybrid rule-based tagger which automatically infers rules from a training corpus based on transformation-based error-driven learning. In my thesis, Brill's tagger is compared to David Arathorn's model of map-seeking circuits that I have adapted for use with natural language processing. I will explain the basics of how a map-seeking circuit works and how I am using it in an attempt to create a Part of Speech tagger that is language independent.