Tokenization is a vital method in natural language processing (NLP), acting as the initial phase for preparing text data. Essentially, it involves splitting up a larger block of text – like a phrase – into smaller units called “segments". These tokens can be terms, punctuation marks, or even sub-