NLP: Finite State Transducer for Morphological Parsing

Share it with your friends Like

Thanks! Share it with your friends!


CS 301 — Spring 2015
Presented by Mike M. and Jenny S.


Maimul says:

gr8 video~ god bless you guys!

Muhammad Hisyam Zayd says:

now i got it, thanks mr

Sanket Gadge says:

Awesome had me 5marks , thanx alot

bharath_bbx says:

Good explanation!

Er Somnath Paul says:

How to create an FST for E-Deletion?

jupiter 6 says:

you mentioned any tape can be output or input, but that is dangerous to say as a closure property of transducers is inversion(reversal to going to input from output string), so we must be specific.

Secondly, if Morph. parsing is breaking down a string to its morphemes and morphological features, then shouldn't the output string i.e 'b' in 'a:b' (input=a)…
Where 'b' is the parsed version of input 'a' = cats.

in words, shouldn't the surface level be the input and lexical level be the output.

This is a serious confusion of mine, as the whole point of morph. parsing is to breakdown(output) a word to its morphologically parsed version.

The text also says that the mapping is done from the surface level(Cats) to the lexical level (cat + noun +pl)

سلمان العنزي says:

thank you, from Saudi Arabia ..

Write a comment


Area 51