#but proofreading for consistency is definitely within the realm of the possible

[liapher]

academic pet peeve: when authors change what variable names refer to halfway through the paper

the authors: let’s look at a sentence! it consists of n words and has a corresponding POS tag sequence that is also of length n. we estimate the joint probability of these two sequences like so: [equation where n is the length of a sentence and the variable i is used to refer to individual words/tags within that sentence]

me: cool, that’s very easy to understand!

the authors in the very next paragraph: so our training set consists of n sentences. we refer to individual sentences using i. the length of sentence i is n_i.

......it would have been so easy to define the training set as a collection of m sentences, indexed with j, where the length of the jth sentence is n_j and each of the words/tags within a given sentence can still be indexed using i.........