COS4861/ 202/0/2020
Natural Language
Processing
Year module
School of Computing
, COS4861/202/0/2020
Tutorial Letter 202/0/2020
Natural Language Processing
COS4861
Year module
School of Computing
This tutorial letter contains model solutions for assignment 02
BAR CODE
Define Tomorrow.
, ASSIGNMENT 2
Solution
Total Marks: 126
Unique Assignment Number: 805059
Please note that it is your responsibility to check that your assignment is registered on the as-
signment database. You can do this by visiting myUnisa.
Chapters covered in this assignment: 5,12.
Question 1: 31 Marks
Visit the universal dependencies project site’s universal POS tags at
https://universaldependencies.org/u/pos/, and review the tag set that is provided there. An-
swer the following questions.
(1.1) Discuss the importance of word classes in natural language processing. (5)
Word classes allow us to differentiate between different words. And differentiating or
classifying different words in a sentence (POS tagging) is the first step in language
processing. Each word has a typical function within a sentence and being able to
classify each word provides clues to the word’s function and ultimately allow us to infer
meaning from a sentence.
{0}
(1.2) Use the following sentence to explain POS tag ambiguity: “We saw her duck”. (5)
Tag the sentence (by hand) using both the Brown and Universal tag-set to illustrate
the problem of ambiguity for each tag-set, and discuss the reason for this ambiguity.
Also compare the possible ambiguity that may arise between the tagsets.
• Brown: We/PPSS saw/VDB her/PP$ duck/(VB or NN)
• UTS: WE/PRON saw/VERB her/DET duck/(NOUN or VERB)
This is plain lexical ambiguity which arises because a word may function as different
types depending on context. ‘Duck’ in this case could be a verb (getting out of the
way), or a noun (the billed water-fowl).
Between the UTS and Brown we see that ‘saw’ in brown is tagged as past-tense,
but no such information is provided in UTS, also ‘her’ in Brown is recognised as a
possessive, while in UTS it is considered a word that modifies a nominal – in the case
duck is considered a NOUN.
{0}
2
Natural Language
Processing
Year module
School of Computing
, COS4861/202/0/2020
Tutorial Letter 202/0/2020
Natural Language Processing
COS4861
Year module
School of Computing
This tutorial letter contains model solutions for assignment 02
BAR CODE
Define Tomorrow.
, ASSIGNMENT 2
Solution
Total Marks: 126
Unique Assignment Number: 805059
Please note that it is your responsibility to check that your assignment is registered on the as-
signment database. You can do this by visiting myUnisa.
Chapters covered in this assignment: 5,12.
Question 1: 31 Marks
Visit the universal dependencies project site’s universal POS tags at
https://universaldependencies.org/u/pos/, and review the tag set that is provided there. An-
swer the following questions.
(1.1) Discuss the importance of word classes in natural language processing. (5)
Word classes allow us to differentiate between different words. And differentiating or
classifying different words in a sentence (POS tagging) is the first step in language
processing. Each word has a typical function within a sentence and being able to
classify each word provides clues to the word’s function and ultimately allow us to infer
meaning from a sentence.
{0}
(1.2) Use the following sentence to explain POS tag ambiguity: “We saw her duck”. (5)
Tag the sentence (by hand) using both the Brown and Universal tag-set to illustrate
the problem of ambiguity for each tag-set, and discuss the reason for this ambiguity.
Also compare the possible ambiguity that may arise between the tagsets.
• Brown: We/PPSS saw/VDB her/PP$ duck/(VB or NN)
• UTS: WE/PRON saw/VERB her/DET duck/(NOUN or VERB)
This is plain lexical ambiguity which arises because a word may function as different
types depending on context. ‘Duck’ in this case could be a verb (getting out of the
way), or a noun (the billed water-fowl).
Between the UTS and Brown we see that ‘saw’ in brown is tagged as past-tense,
but no such information is provided in UTS, also ‘her’ in Brown is recognised as a
possessive, while in UTS it is considered a word that modifies a nominal – in the case
duck is considered a NOUN.
{0}
2
