Starting and Ending Sentence Lecture - End of sentence, typo?

Course Q&A Natural Language Processing NLP with Probabilistic Models
1

In the case of ‘small corpus’ with only 3 sentences, made up of 2 words yes and no.

At 3:30, it is said that formulation is shown on screen is for bigram probability. But isn’ it a trigram probability ?

P(<s> yes yes) is trigram probability notation, considering the start token right ?
If start token is not considered as an element then why is it equated/expanded to P(yes/<s>) x P(yes/yes) ?

image
image1147×397 49 KB

2

Hi salih-g,

The formulation serves to calculate the bigram probability of the sentence, i.e. using bigrams to calculate the probability of the overall sentence. In the given example, this probability consists of the bigram of the word yes following the start token times the bigram of the word yes following the word yes.

Maybe this exposition clarifies further.