Stanford CoreNLP

People not infrequently complain that Stanford CoreNLP is slow or takes a ton of memory. In some configurations this is true. In other configurations, this is not true. This section tries to help you understand what you can or can't do about speed and memory usage. The advice applies regardless of whether you are running CoreNLP from the command-line, from the Java API, from the web service, or from other languages.