Language model

Language modelling is a practice within artificial intelligence whereby a computer is trained to
recognise – and to a limited extent interpret – words by a probabilistic analysis of a large body of text in a given language. Once the corpus has been fed into a computer, a bespoke script will run that determines the overall number of instances of a particular word and also the words that
it is most/least often associated with in a phrase. Language modelling can thus be used to
refine the OCR process, by excluding those words that are statistically least likely to be associated with one another.