On lecture, Interpolation was explained in the context of different types of smoothing, for the case where an N-gram is missing from the corpus.

Can we (or even should we) use interpolation also as a general technique for estimating probabilities, regardless of whether the N-gram is missing or not? That is, can we use interpolation even if N-gram is *not* missing?

hi @Doron_Modan

interpolation is a technique used in natural language processing (NLP) to estimate a value between two known points in a data set

it’s another smoothing technique to combine context (corpus) from multiple types of N-gram into the computation of the probabilities.

I don’t the lecture mentioned Interpolation is used when Ngram is missing from the corpus, rather it states interpolation is used when if an N-gram is missing, like you have unigram and bigram to say, but you do not have trigram, so at this time interpolation helps to find a value by lower the n gram value by 1, so standard smooth curve can be estimated based on the probabilities

check the below image