8 thoughts on “Computational Linguistics I: Morphology”
Thanks for this great lecture!
One question concerning non-deterministic FSAs: Why do you need the non-deterministic start at all? Can't you go to q3, q6, q15 and q17 immediately from the start state?
Yes, you can. In this case, however, the nondeterminism is a convenience. To glue multiple FSAs together, you can just say: I now have an epsilon transition to all of the start stats of the FSAs I'm gluing together.
It allows yourself not to think about the specific construction of the FSAs.
I see, thanks. Are there also cases where non-determinism is actually necessary for the FSA to work?
Nope. You can always make a non-deterministic FSA deterministic. Non-determinism is just a notational convenience.
How common are courses in English about morphological parsing of other languages like the agglutinative Japanese or morphologically-simple Chinese? I'm just curious if any Anglophone university teaches computational linguistics targeting another language, or if that's just too impractical.
First of all thanks for this informative lecture sets. In slide number 7, Skyped, Faxed are mentioned as derivational morphology. Shouldn't they be inflectional or hybrid? Skype is a noun which is changed to verb and then it changed into the past form of that verb…
good work
Thank You for excellent explanation,and scientific view point on computational linguistics.Lots of my doubts cleared,and inspired for higher research in computational linguistics.Dr. Amit Purushottam
Thanks for this great lecture!
One question concerning non-deterministic FSAs: Why do you need the non-deterministic start at all? Can't you go to q3, q6, q15 and q17 immediately from the start state?
Yes, you can. In this case, however, the nondeterminism is a convenience. To glue multiple FSAs together, you can just say: I now have an epsilon transition to all of the start stats of the FSAs I'm gluing together.
It allows yourself not to think about the specific construction of the FSAs.
I see, thanks. Are there also cases where non-determinism is actually necessary for the FSA to work?
Nope. You can always make a non-deterministic FSA deterministic. Non-determinism is just a notational convenience.
How common are courses in English about morphological parsing of other languages like the agglutinative Japanese or morphologically-simple Chinese? I'm just curious if any Anglophone university teaches computational linguistics targeting another language, or if that's just too impractical.
First of all thanks for this informative lecture sets. In slide number 7, Skyped, Faxed are mentioned as derivational morphology. Shouldn't they be inflectional or hybrid? Skype is a noun which is changed to verb and then it changed into the past form of that verb…
good work
Thank You for excellent explanation,and scientific view point on computational linguistics.Lots of my doubts cleared,and inspired for higher research in computational linguistics.Dr. Amit Purushottam