Xrenner to JSON-NLP

(C) 2019 by Damir Cavar, Oren Baldinger, Maanvitha Gongalla, Anurag Kumar, Murali Kammili, Boli Fang

Brought to you by the NLP-Lab.org!

Introduction

Xrenner wrapper for JSON-NLP. Xrenner specializes in coreference and anaphora resolution, in a more highly annotated manner than just a coreference chain.

Required Dependency Parse

Xrenner requires a Dependency Parse in CoNLL-U format. This can come from CoreNLP, or another parser that provides universal dependencies in [CoNNL-U] format. There are two ways to accomplish this:

CoreNLP Server

The XrennerPipeline class will take care of the details, however it requires an available CoreNLP server. The easiest way to create one is with Docker:

docker pull nlpbox/corenlp
docker run -p 9000:9000 -ti nlpbox/corenlp

To test this, open a new tab,

wget -q --post-data "Although they didn't like it, they accepted the offer."   'localhost:9000/?properties={"annotators":"depparse","outputFormat":"conll"}' -O /dev/stdout

You then need to create a .env file in the root of the project, follow the example in sample_env. The default entry that corresponds to the Docker command above is:

CORENLP_SERVER=http://localhost:9000

Provide your own CoNLL-U

Use the XrennerPipeline.process_conll function, with your conll data passed as a string via the conll argument.

You may find the pyjsonnlp.conversion.to_conllu function helpful for converting JSON-NLP, maybe from spaCy, to CoNLL-U.

Microservice

The JSON-NLP repository provides a Microservice class, with a pre-built implementation of Flask. To run it, execute:

python xrennerjsonnlp/server.py

Since server.py extends the Flask app, a WSGI file would contain:

from xrennerjsonnlp.server import app as application

Text is provided to the microservice with the text parameter, via either GET or POST. If you pass url as a parameter, the microservice will scrape that url and process the text of the website.

Here is an example GET call:

http://localhost:5000?text=John went to the store. He bought some milk.

The process_conll endpoint mentioned above is available at the /process_conll URI. Instead of passing text, pass conll. A POST operation will be easier than GET in this situation.