Cource: Natural Language Processing

1b) Find the lexical translation probabilities of the foreign words a, b, c, d, e from the following bilingual corpora using the statistical approach. Also find the best translation for the foreign sentence, cde calculating the lexical translation probability in IBM Model-I

Foreign Sentences	Target Sentences
a b c d e	X Y Z P Q
a b c c d	X Y Z Z R
a c d d b	M K R P N
a e e b e	M Q O Y T

1c) Show an alignment function for the following cases where the source and target languages are English and Bengali, respectively.

Reordering.

One to many.

Dropping words.

Inserting words.

3c) Using the Hash Trick insert the words BOY, TOY and JOY in the hash table of size 4.

4a) A hearing-impaired person in Bangladesh cannot hear sound, but he/she wants to understand the sound- What is the automated mechanism to understand the speech? Explain with a diagram.

4b) A blind person in a rural area of Bangladesh wants to read text, but he/she cannot read-What is the automated mechanism to understand the text? Explain with a diagram.

4c) Find the output utterance for the test corpus given below using the Euclidean distance

Training Corpus: Test Corpus

/a/:	2	5 6	3 5 4
/i/:	3	4 5	2 3 3
/i/:	5	2 4
/a/:	4	5 3
/a/:	4	6 3