Introduction

This repository contains scripts to create transformer-based deep learning models for the prepositional phrase attachment problem, as described in:

Adwait Ratnaparkhi and Atul Kumar. (2021). Resolving Prepositional Phrase Attachment Ambiguities with Contextualized Word Embeddings. In Proceedings of ICON-2021: 18th International Conference on Natural Language Processing. National Institute of Technology, Silchar, India. December 16-19, 2021

Building the data

There are two data prepositional phrase attachments sets used by this work:

• Binary classification
  • This is the data set from (Ratnaparkhi et al, 1994).
  • It is located here.
• Multi-way classification
  • This data set is derived from the scripts and data in (Belinkov et al, 2014).
  • The scripts to build it are here

Setup

conda create -n ppa python=3.8

conda activate ppa

# from pytorch.org
conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch 

pip install transformers==4.2.1

pip install datasets

Training and evaluating the models

The following shell script will reproduce the experiments in the above paper. It will call the scripts in various sub-directories. Make sure you have built the BLBG data before calling this script!

cd scripts
./run_all.sh

Results will be stored in a directory named results_XXXXXXX in the sub-directory, in a file named test_results.txt

References

[1] Yonatan Belinkov, Tao Lei, Regina Barzilay, and Amir Globerson. (2014). Exploring compositional architectures and word vector representations for prepositional phrase attachment. Transactions of the Association for Computational Linguistics, 2:561–572.

[2] Adwait Ratnaparkhi, Jeff Reynar, and Salim Roukos. (1994). A maximum entropy model for prepositional phrase attachment. In HUMAN LANGUAGE TECHNOLOGY: Proceedings of a Workshop held at Plainsboro, New Jersey, March 8-11, 1994.