Tokenization is a crucial step in natural language processing (NLP), acting as the initial phase for preparing text data. Essentially, it involves breaking down a larger block of information – like a sentence – into smaller units called “segments". These tokens can be copyright, punctuation marks