CS 124 PA 6: Direct Machine Translation

Due February 28th @ 5:00PM To Do + Deadlines:

By Saturday (1-4 complete):

1. Choose a language

2. Build a test corpus for that language (15 sentences from outside sources: save sources); pick 10 sentences for the dev set and 5 for the test set

3. Create a dictionary for all the words in the corpus using www.wordreference.com or google translate.

• If there’s more than one definition, include all of them in the corpus and we’ll come up with good heuristics for choosing the correct one

4. Look up annotation toolkits for the chosen language and determine how to use them

5. Actual coding:

i) Use the dictionary to translate the dev corpus into English words and find annotation tools for sentences that will be helpful for post processing

ii) Come up with 6-10 post processing strategies to improve the baseline translations and code them

6. Run the system on the test set

7. Error analysis

8. Compare our result with google’s

9. Follow up report