4
votes
$\begingroup$

I have tried open nlp NER for extracting organization names with not great success (It could be model is not fit for the domain I am working). So, I am planning to train Open NLP NER on my training data.

The point I am stuck on is, Annotating text. I searched on internet and found that GATE, Knowtator, BART etc., but they are complicated to setup on location machine, as well as I am not sure which tools are being used by Open NLP community for this.

Could someone please help me pointing to good Annotation tool and any references on how to use them for this purpose?

$\endgroup$
  • 3
    $\begingroup$ Personally, I've just used Excel for this. E.g. Put the sentence word by word in one row, one word/toekn per column, and then little drop-down boxes with valid values or free text input in the row(s) underneath. Then write a python script with openpyxl to read/write the excel files. It's ugly (like anything with excel involved), but it works - most important is that you don't have to train people in how to use Excel, and most IT departments have to support Excel so the pushback you get from "install Excel on this machine" is zero. $\endgroup$ – Patrick Caldon Aug 7 '13 at 2:33
  • $\begingroup$ @PatrickCaldon: Thanks for your comment. Let us see if any other valuable answers from community members. $\endgroup$ – kosa Aug 7 '13 at 3:14
  • $\begingroup$ Can you extend your "etc"? $\endgroup$ – Franck Dernoncourt Jun 6 '15 at 21:10
3
votes
$\begingroup$

Try http://nlp.stanford.edu/software/stanford-manual-annotation-tool-2004-05-16.tar.gz

It's very simple to use. Just launch the jar file, create your tags, and start annotating. It even has a batch mode so you can keep annotating file by file. You point the app to a folder of text input files and it outputs the annotated files to a different folder.

$\endgroup$
2
votes
$\begingroup$

I used the Knowtator plugin for Stanford's Protégé ontology editor. I used throughout the text annotation process for my PhD dissertation, which involved annotating text for training classifiers. Although it's not the simplest to use, it's definitely the best tool I could find after surveying the list of possible software.

$\endgroup$
1
vote
$\begingroup$

You can use Dataturks, it gives you simple UI with collaborative enviroment so that it takes minutes for any data science/machine learning team to create annontated data. Its easy to invite people to tag using emails. Give you dynamic insights based on usecase. Supports complete annotation usecases.

https://dataturks.com

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.