LNMIIT #
LNMIIT/B. Tech. /CSE/PE/2019-20/ODD/CSE3201/ET
The LNM Institute of Information Technology
Department of Computer Science & Engineering
‘CSE 3201 Natural Language Processing
Exam ‘Type: End ‘Term
‘Timer 180 min 10/12/2019, Max. Marks: 50
Answer all questions in the same order as it appears in the question paper. If there are any
‘assumptions to be made for your answer write clearly the assumption that you are making before
‘answering, Only if the assumption is reasonable it will be considered. No doubt clarifications in the
examination hall! All the best!
1, Give an example for zeugma. 2)
2, Develop a set of regular expressions to recognize each one of the character shape features (separately)
Capitalized, All caps, Mixed caps, and Ends in digit @
4. Use the Viterbi algorithm and the given HMM below to compute the most likely weather se-
quences for each of the Ovo observation sequences 331122313 and 311123312. Draw the Trellis and
6)
fil) up the Trellis by using, Viterb's algorithm
ne[.8..2] : oO O :
.
abl
4. Names of works of art (books, movies, video games, ete.) are quite different from the kinds of named
tntities that we have discussed in the class, Let us suppose that we have a collection containing a
list of names of works of art from a particular category from a web-based source (eg. gutenberg. org,
amazon.com, imdb.com, ete.). Analyze this list and give examples of ways that Uke names in it are
likely 10 be problematic for the techniques described for Named Entity Recognition. 6)
Consider the following queries:
Who did the Vice President ball?
Who billed the former Treasury Secretary?
(a) What type of queries are these? Substantiate your argument with only one sentence. @)
(b) Suppose you pose these queries to your favorite Web search engine will it give appropriate answers?
‘What are the the issues i will have? How these issues can be addressed by a more intelligent
question-answering system? Do not write it a8 a paragraph enumerate your points one by one
6)
Al the best!|"2
| a
LNMIIT/D. Teck /OSE/PE/2019-0/OD0/CSE201/E1 LNMIIT
SE sie “pine cou. Using the La agar [eto Asanbiguate the pres
‘The following dictionary definitions are used:
PINE
1. kinds of evergreen tree with ueedle-shaped leaves:
2, waste away through sorrow or sles
3, to lose vigor, health, or flesh (as through grief)
a, to strongly desire especially something that is difficult or lost for
CONE
1. solid body which narrows to @ point
2, something of this shape whether solid or hollow
3. fruit of certain evergreen trees
Explain the method used clearly. Suppose we use the word “pine” asin thi sentence: “Dan is sll
pruning for hi lost dog”, how does the Lesk algorithm work to citambiguate the word *pining!? Fxp}as
the steps involved very briefly. ata)
1. Using Farley Parsing algorithin parse the sentence “Take that bottle”. List all the steps involved by
follosing the chart method described in the class. Use the following sot of rules for the parsing. (10)
S—NPVP Nominal —> Nominal Noun
S—> Aus NP VP Nominal —> Nominal PP
S—>VP VP Verb
NP —> Pronoun VP — Verb NP
NP —+ Proper-Noun VP —+Verb NP PP
NP — Det Nominal VP — Verb PP
Nominal —+ Noun VP —+VP PP
Noun —> bottle PP —+ Preposition NP
Det => that Verb —» Take
8. In Good-Turing Discounting we know that c” = (c-+1) gh which gives the count forall bins N)..No
We avo know that a
n(vhings with frequency zero in trating) = 5
Prove that ¢* derives the value for Pér(things with frequency zero in training) (that isthe latter follows
from the former), You may make all necessary assumptions. (6)
‘All the best!