Masters Thesis

Automated part of speech induction to improve understanding of the parts of speech

While the concept of parts of speech is often treated as being well understood, a review of an assortment of applications shows imperfect consistency and little or no justification for the conventions used. I suggest that automated part of speech induction could be used to improve understanding of the parts of speech. I discuss an established algorithm for part of speech induction and four papers that used that algorithm. I then explain how I modified that algorithm to more clearly reflect the intuitive idea that words that occur in similar contexts should be grouped together. I discuss my implementation of the resulting algorithm and the results of testing it on the Brown Corpus. Finally, I discuss what appear to be the main problems with my algorithm and possible basis for solutions to these problems.

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.