CheNER: Chemical Named Entity Recognizer

CheNER is a named entity recognition tool, that uses Conditional Random Fields for identifying mentions of chemicals in text, focusing on IUPAC entities.

Downloads: CheNER [JAR File] (52M), user documentation [PDF].

Usage:

To run the user interface:

• from the command-line execute: java -jar CheNER.jar
• double-click on the jar file (in Windows operating system)
  

To run an example of how to tag a simple string text, please execute from the command-line:

java -cp paht/to/CheNER.jar Examples.TaggingExample

To run an example of how to train a customazed dataset, please execute from the command-line:

1. java -cp path/to/CheNER.jar Examples.TrainingExample 'trainFile' 'modelFile'
2. java -cp path/to/CheNER.jar Examples.TrainingExample 'trainFile' 'modelFile' 'p1,p2,p3,p4,p5,p6,p7,p8'

   where p1 indicates if MF feature class is activated (1) or not (0), is applied the same logic for the rest of pN in order for ToC, PS, W, WB, L, BNS and SW

To use CheNER in your code:

String inputFile="Glutamic acid analogs containing 3- and 4-methyl and 2-, 3-,"
               + "and 4-phenyl substituents were prepared.";
Tagger t = new Tagger();
//type=0 -> identify IUPAC type entities
//type=1 -> identify All type entities
t.doTheTagging(inputFile, "annotatedText.txt", 0, false);
Hashtable entities=t.getEntities();
Enumeration en=entities.keys();
  while(en.hasMoreElements()){
  String entity=en.nextElement().toString();
  int n=Integer.parseInt(entities.get(entity).toString());
  System.out.println(entity+" ("+n+")");
  }
}

To use CheNER from shell/commandLine:

> java -cp CheNER.jar chener.CommandLineClass

Class Summary:

Tagger Class Detail:

This class uses the CRF interface to do the named entity tagging.
By default all methods use a tokenization by spaces. The tokenization can't be modified.

Trainer Class Detail:

This class will train the CRF to extract the entities from the customized dataset.
The input file or training file must be tokenized with one word per line, with a \tab separating the word/token from its label.The first token of an entity name should have a label "B", all other entity token labels should be "I", and non-entity tokens should be labeled with "O", following the IOB scheme. If the input file contains more than one document, each document must be separatet by a new white line.

The O
synthesis O
of O
a O
range O
of O
3-hydroxy-4(1H)-pyridinones B