# GENIA POS patterns for biomedical term extraction.
# Ported from Java JATE 2.0 (testdata/solr-testbed/GENIA/conf/genia.patterns).
# Original source: http://users.cs.cf.ac.uk/I.Spasic/flexiterm/
#
# Tag type: Universal POS (spaCy pos_)
#
# Translated from Penn Treebank tags:
#   NN/NNS -> NOUN, NNP -> PROPN, JJ -> ADJ, IN -> ADP
#
# Multiple patterns are combined with | (OR) into a single regex.
# Each line is a separate pattern (blank lines and # comments ignored).
# All patterns are applied to a space-separated POS string with trailing space.

# Basic NP: (ADJ|NOUN|PROPN)+ (NOUN|PROPN)
(ADJ |NOUN |PROPN )*(NOUN |PROPN )

# NP with optional preposition: (ADJ|NOUN|PROPN)* ADP? (ADJ|NOUN|PROPN)* (NOUN|PROPN)
(ADJ |NOUN |PROPN )*ADP (ADJ |NOUN |PROPN )*(NOUN |PROPN )
