Question about example - bigram (slide 34)

Course Q&A Natural Language Processing NLP with Probabilistic Models
1 
Hi,
On slide 34 of C2_W3.pdf there is an example:
Corpus
<s> Lyn drinks chocolate </s>
<s> John drinks tea </s>
<s> Lyn eats chocolate </s>
P(John | <s>) = 1/3
P(chocolate | eats) = 1/2
P(</s> | tea) = 1/1
P(Lyn | <s>) = 2/3

My question is about probabilities calculation. On slide 11 there is a formula:
P(y | x) = C(x y) / sum_w C(x w) = C(x y) / C(x). Using it we have:

P(John | <s>) = C(<s> John) / (C(<s> Lyn) + C(<s> John)) = 1 / (2 + 1) = 1/3 (as on the slide)
P(chocolate | eats) = C(eats chocolate) / (C(eats chocolate)) = 1 / 1 (different)
P(</s> | tea) = C(tea </s>) / (C(tea </s>)) = 1 / 1 (as on the slide)
P(Lyn | <s>) = C(<s> Lyn) / (C(<s> Lyn) + C(<s> John)) = 2 / (2 + 1) = 2 / 3 (as on the slide)

Could you please explain why the slide says that P(chocolate | eats) = 1/2?
2

Hi @conscell

You found a mistake. The slide will be fixed. :+1:

Since there is only one “eats” (followed by “chocolate”), the probability for P(chocolate | eats) = 1.

Thanks.

3

Hi @arvyzukai,
Thank you very much for clarifying this.