Node-NER (Named Entity Recognition)

Node-NER uses Stanford's JAVA NER package to tag the entities in the text, then parse the output to extract the entities by type.

Installation

Dependency

First, you will need to download Stanford NER.

Direct link

Download page

Unzip anywhere, it doesn't matter.

Note: Don't forget to have your JRE up to date, and JAVA in your PATH.

You can make sure everything is set by opening a console, cd into the stanford-ner's directory then execute: java -mx1500m -cp stanford-ner.jar edu.stanford.nlp.ie.crf.CRFClassifier -loadClassifier classifiers\english.all.3class.distsim.crf.ser.gz -textFile path-to-a-text-file.txt

NPM Package

npm install node-ner

Code Example

var node_ner = require('node-ner');

var ner = new node_ner({
	install_path:	'/path/to/stanford-ner'
});

ner.fromFile('/path/to/a/file.txt', function(entities) {
	console.log(entities);

	/*
	Output:
	{
		"ORGANIZATION": [
			"Samsung Electronics Co Ltd",
			"Microsoft",
			"Apple"
		],
		"DATE": [
			"Thursday"
		],
		"MONEY": [
			"$ 2 billion"
		]
	}
	*/
})

Limitations

You can only load from a file at this moment. This package is under development, syntax and specs can change at any time.

License

http://www.gnu.org/licenses/gpl-2.0.html