edit config file in stanford pos tagger

Question

i have tagged a simple sentence and this is my code:

package tagger;

import edu.stanford.nlp.tagger.maxent.MaxentTagger;

public class myTag {

public static void main(String[] args) {

    MaxentTagger tagger = new MaxentTagger("D:/tagger/english-bidirectional-distsim.tagger");


    String sample = "i go to school by bus";

    String tagged = tagger.tagString(sample);

    System.out.println(tagged);
}

}

this is the output:

    Reading POS tagger model from D:/tagger/english-bidirectional-distsim.tagger    ... done [3.0 sec].
i_LS go_VB to_TO school_NN by_IN bus_NN

after editing the properties file it doesn't have any effect at all. for example i have changed the tag separator to ( * ) but in the output it still prints ( _ ).

how could i use the model config file in eclipse?

Note that the bidirectional model [is not generally the best model to use in real-world applications](http://nlp.stanford.edu/software/pos-tagger-faq.shtml#h). There are other models that sacrifice a small amount of accuracy for a substantial increase in speed. — Ian Roberts, Apr 03 '15 at 12:46

Nikita Astrakhantsev · Accepted Answer · 2015-04-03T12:44:39.647

1

You can load Properties file and pass it to the constructor of MaxEnt, something like this:

Properties props = new Properties();
props.load(new FileReader("path/to/properties"));
MaxentTagger tagger = new MaxentTagger("D:/tagger/english-bidirectional-distsim.tagger", props);

You can also set properties in props object directly:

props.setProperty("tagSeparator", "*");

NB: if you use the original properties file and it fails with exception like

java.io.FileNotFoundException: /u/nl
p/data/pos_tags_are_useless/egw4-reut.512.clusters (No such file or directory)

then remove arch and trainFile attributes.

edited Apr 03 '15 at 12:44

answered Apr 03 '15 at 09:57

Nikita Astrakhantsev

4,701
1
15
26

could you please explain more about the first solution>>VM arguments? – Chelsea-fc Apr 03 '15 at 10:27
What is it exactly to be explained? – Nikita Astrakhantsev Apr 03 '15 at 10:53
the second solution works fine but i want to change multiple parameters in properties file so in this case the first solution is better.the question is how to pass the properties file as arguments to program? – Chelsea-fc Apr 03 '15 at 11:03
I've edited sample code (and checked it); try it now. – Nikita Astrakhantsev Apr 03 '15 at 12:45
I think you should create another question and explain your problem in detail. In particular, what is the expected result, the actual one (seems to be presented here) and steps-to-reproduce. Currently I don't get your problem. – Nikita Astrakhantsev Apr 03 '15 at 21:27
this is the new question,tahnks.[link] (http://stackoverflow.com/questions/29443556/xml-format-in-stanford-pos-tagger) – Chelsea-fc Apr 04 '15 at 06:44

Ganesh · Answer 2 · 2016-06-04T13:55:46.690

Instead of writing a java code for this, you can use the bash file which comes in the downloaded ZIP file. After extracting the postagger's ZIP file, edit the following bash file:

stanford-postagger.sh

It should have the following line:

java -mx300m -cp 'stanford-postagger.jar:lib/*' edu.stanford.nlp.tagger.maxent.MaxentTagger -model $1 -textFile $2

Add a parameter called "-tagSeparator [YourTag]" after "-model $1":

java -mx300m -cp 'stanford-postagger.jar:lib/*' edu.stanford.nlp.tagger.maxent.MaxentTagger -model $1 -tagSeparator * -textFile $2

To run it (Make sure necessary permissions are given):

./stanford-postagger.sh models/model_name.tagger in_filename > out_filename

Voilà!

edit config file in stanford pos tagger

2 Answers2

Linked