CoreNLP

Why we included this project

Stanford CoreNLP is one of the oldest integrated NLP toolkits written in Java, and it still holds up as a single library that covers the full text-analysis pipeline. Feed it raw text and it returns tokens, part-of-speech tags, named entities, normalized dates and numbers, dependency parses, and coreference links through one API. That appeals to teams building extraction or search features in Java who want stable, well-tested components without juggling a pile of dependencies. It ships pre-trained models for several languages besides English, among them Arabic, Chinese, French, German, and Spanish, so you can do multilingual work without training your own. The underlying tools mix rule-based, statistical, and neural methods, which gives predictable results on classic NLP tasks rather than a scramble to stay on top of the latest benchmarks.