I have been trying hard to understand the concept of sampling some contexts. The main thing is I am unable to digest the idea of sampling. For example in Mikolov's papers, the negative sampling expectation is formulated as : logσ(⟨w,c⟩)+k⋅𝔼cN∼PD[logσ(−⟨w,cN⟩)] I clearly understand the left term logσ(⟨w,c⟩), but I can't understand the very idea of sampling negative word-context pairs.