In natural language processing, lemmatisation is a key pre-processing step that reduces words down to their root word (i.e., conjugated verb to the present tense, plurals to the singular). The idea behind this is that these shared-root words have the same value to our analysis, so we want a single, consistent representation in our corpus.