Dear INLP students,
Now the quiz 2 grades shall be ok.
Below is the global distribution.
Two main remarks about this quiz:
- many (2/3 of the answers) claimed that "Pos tagging aims to suppress lexical variability of words", which is wrong; this is the aim of the lemmatization; it seems that when reading a slide some forget to read its title and/or it's bottom :-(
- the most discriminative (and failed) question was about HMM parameters (number): remember there are
- initial probabilities (= tagset size)
- transition probabilitities (= tagset size to the power the order of the HMM plus one)
- emission probabilities (= vocabulary size times tagset size)
I also computed "your fraction" so far (average of the two quizzes, in percent of the total points) and the corresponding grade it would be if we would stop here (but remember this is only 8% of the final grade so far).
Best,
