Final Project Topic Suggestions

1. Come up with your own idea!
2. Try one of the SensEval competitions' tasks. The good thing about these tasks is that the test and training data are all there. The bad thing is that we won't cover these topics until after the midterm.
3. Expand assignments 3 and 4 with a grammar and lexicon for your choice of language (other than English).
4. Expand assignments 3 and 4 with a grammar and lexicon so that it can serve as a NL front end for your choice of program.
5. Implement a simple natural language generator.
6. Expand assignments 3 and 4 to handle probabilistic context-free grammars. Or if you have a group of students, probabilistic lexicalized CFGs.
7. Train a statistical machine translation model using the CRATER corpus from assignment 2 and compare how well your program does on similar topic web pages versus Google Translate. Unfortunately we won't cover machine translation until near the end of the semester, but most of the techniques (HMMs and EM) are very similar to what we've covered already.

Email me a short description of your project idea for feedback.