NP Chunking Demo

This is a demonstration of an NP chunker, a program that recognizes base noun phrases in a text. This chunker is relatively small in order to make it faster. It uses the Memory-Based Learning algorithm IB1-IG and it has been trained on section 15 of the Wall Street Journal corpus (50442 words). It finds approximately 89% of the baseNPs in a text and of the NPs it extracts from the text approximately 89% is correct. For more elaborate NP chunking programs these figures lie around 92%.

Related information: text on NP chunking.


Enter a text here with punctuation marks separated from the words by white space and press submit. Processing will take about 30 seconds.


Last update: November 08, 1999. erikt@uia.ua.ac.be