Accuracy

spencer kelly edited this page Jul 5, 2017 · 5 revisions

nlp_compromise was built to sacrifice a couple percentage-points on accuracy, in order to accomplish goals in filesize, ease-of-use, and maintain the playful pragmatism of the javascript demoscene.

While accuracy is not the most important metric, the performance tests can be run per-build, and a mid-80's precision is to be expected in most tasks.

because nlp_compromise makes independent decisions about term & sentence lumping/splitting, evaluating the accuracy of POS-tagging accuracy against a corpus is non-trivial but expects results between 84%-86% on Penn.

You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.