textacy

Why we included this project

textacy sits on top of spaCy and handles the work that spaCy itself does not bother with. It gives you utilities for cleaning and normalizing raw text before parsing, and after processing you can extract n-grams, entities, acronyms, keyterms, and subject-verb-object triples without writing the extraction code yourself. Tokenization and vectorization helpers, topic modeling, string similarity comparisons, and readability statistics such as Flesch-Kincaid grade level and type-token ratio round out the library. That makes it a solid pick for data scientists and NLP engineers building analysis scripts or research prototypes where spaCy does the parsing and textacy does the rest. Bundled datasets covering Congressional speeches, historical literature, and Reddit comments are handy for trying ideas quickly instead of hunting down corpora on your own.