Can a machine think? Can an algorithm grasp a literary concept? Based on a formal definition, we are building and training algorithms capable of identifying and understanding a literary concept through texts. As part of the IAL project, we are working in particular on the notion of ‘variation’ in the Greek Anthology: identification of variations (repetitions, rewritings, reformulations) in the epigrams, formal definition of the concept, implementation of the definition in an algorithm. The algorithm, capable of identifying without fail all the variations previously identified as such, will embody our definition and ‘understand’ the concept of variation.