This page contains the code used in the work "Learning with Structured Representations for Negation Scope Extraction" published at ACL 2018.
Prerequisite: JRE (1.8 or later)
Run the following command to try out the Linear, Semi and Latent models described in the paper.
java -Xmx32g -Xms8g -jar ns.jar -modelname OIBAMN6 -thread 4 -subpath default -outputscope true -num_iter 1000 -reg 0.1 -optimizer lbfgs -dataset cdsco -lang en -discrete true -discardintest false -syntax true -useperl true -outputsem2012 true -unipos true
java -Xmx32g -Xms8g -jar ns.jar -modelname SEMIBOI2 -thread 4 -subpath default -outputscope true -num_iter 1000 -reg 0.1 -optimizer lbfgs -dataset cdsco -lang en -discrete true -discardintest false -syntax true -useperl true -outputsem2012 true -unipos true
java -Xmx32g -Xms8g -jar ns.jar -modelname OIBAMNLatent6 -thread 4 -subpath default -outputscope true -num_iter 1000 -reg 0.1 -optimizer lbfgs -dataset cdsco -lang en -discrete true -discardintest false -syntax true -useperl true -outputsem2012 true -unipos true -latentmax 2
Or you can simply run
./exp_cdsco.sh
Please download the Bioscope and CNeSp dataset before trying out the additional experiments. Use the following scripts for running such additional experiments:
./exp_bioscope.sh
./exp_cnesp.sh
The source code is written in Java, which can be found under the "src" folder.
Note that it is a Maven project, therefore we recommend you to import this project as a Maven project in eclipse.
If you use this software for research, please cite our paper as follows:
@InProceedings{P18-2085,
author = "Li, Hao and Lu, Wei",
title = "Learning with Structured Representations for Negation Scope Extraction",
booktitle = "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
year = "2018",
publisher = "Association for Computational Linguistics",
pages = "533--539",
location = "Melbourne, Australia",
url = "http://aclweb.org/anthology/P18-2085" }
Email to [email protected] if any inquery.