1. Write a method to perform Viterbi decoding to find the best sequence of tags for a line (sequence of words).
2. Test the method on simple hard-coded graphs and input strings. In your report, discuss the tests and how they convinced you of your
3. Write a method to train a model (observation and transition probabilities) on corresponding lines (sentence and tags) from a pair of
4. Write a console-based test method to give the tags from an input line.
5. Write a file-based test method to evaluate the performance on a pair of test files (corresponding lines with sentences and tags).
6. Train and test with the two example cases provided in [url removed, login to view]: "simple", just some made up sentences and tags, and "brown", the
full corpus of sentences and tags. The sample solution got 32 tags right and 5 wrong for simple, and 35109 right vs. 1285 wrong for brown, with an unseen-word penalty of -100. In a short report, provide some example new sentences that are tagged as expected and some that aren't, discussing why. Also discuss your overall testing performance, and how it depends on the unseen-word penalty (and any other parameters you use).
4 freelancers are bidding on average $64 for this job
I'm a computer science professional with PhD degree and extensive experience in Java, C, C++, C#, Python, Perl, PHP, databases, software engineering. It would be my pleasure to complete your project.
Hello, I am really interested to work on this project, I can start the work now , and can provide the best services from my end. Please come on chat to discuss more about the project. Thanks!