A NLP tool that will be used by OnAIR project. It reads an input file formatted such as each line represents a sentence and outputs its analysis.
The core NLP library used is Apache OpenNLP, wrapped by CoGrOO API, as well as some CoGrOO tools, like the Featurizer and Lemmatizer. English models are provided by OpenNLP, while Portuguese models are provided by CoGrOO.
Download onair-nlp.jar from the download panel.
Usage: java -jar onair-nlp.jar lang inputFile outputFile encoding
supported languages: pt en
To analyze an UTF-8 English document: java -jar onair-nlp.jar en in.txt out.txt UTF-8
To analyze an UTF-8 Portuguese document: java -jar onair-nlp.jar pt in.txt out.txt UTF-8
Column 1: lexeme
Column 2: lemma (Portuguese only)
Column 3: POS tag
Column 4: Features (Portuguese only)
Column 5: Chunks
Fantasia [fantasia] n F=S B-NP
é [ser] v-fin PR=3S=IND B-VP
o [o] art M=S B-NP
nome [nome] n M=S I-NP
dado [dar] v-pcp M=S B-VP
a [a] prp - B-PP
diversos [diverso] adj M=P B-NP
espetáculos [espetáculo] n M=P I-NP
equestres [equestre] adj M=P I-NP
tradicionais [tradicional] adj M=P I-NP
que [que] pron-indp M=P B-NP
simulam [simular] v-fin PR=3P=IND B-VP
assaltos [assalto] n M=P B-NP
militares [militar] adj M=P I-NP
The tagset is derived from Floresta tag set Check the sections 2.1 (Group forms) and 2.3 (Word classes).
OnAIR-NLP is published under the Apache License 2.0, expect for the English models that are not suitable for commercial use. Please refer to Apache OpenNLP downloads documentation for details on this.