Skip to content

Latest commit

 

History

History
21 lines (13 loc) · 702 Bytes

chunking.md

File metadata and controls

21 lines (13 loc) · 702 Bytes

Chunking

Chunking is a shallow form of parsing that identifies continuous spans of tokens that form syntactic units such as noun phrases or verb phrases.

Example:

Vinken , 61 years old
B-NLP I-NP I-NP I-NP I-NP

Penn Treebank

The Penn Treebank is typically used for evaluating chunking. Sections 15-18 are used for training, section 19 for development, and and section 20 for testing. Models are evaluated based on F1.

{% include table.html results=site.data.chunking scores='F1 score' %}

{% include chart.html results=site.data.chunking score='F1 score' %}

Go back to the README