PathNet: Exploiting Explicit Paths for Multi-hop Reading Comprehension

This repository contains the source code for the paper Exploiting Explicit Paths for Multi-hop Reading Comprehension. This work was published at Association of Computational Linguistics (ACL) 2019. If you find the paper or this repository helpful in your work, please use the following citation:

@inproceedings{pathnet,
  title={ Exploiting Explicit Paths for Multi-hop Reading Comprehension },
  author={ Souvik Kundu and Tushar Khot and Ashish Sabharwal and Peter Clark },
  booktitle={ ACL },
  year={ 2019 }
}

Setup

We used Python-3.6.2. Consider creating a virtual/conda environment for development. This code repository is built using AllenNLP. To install all dependencies please run the following:

sh scripts/install_requirements.sh

Download

To download all the required files, run scripts/download.sh

Prediction for WikiHop

Once you run the scripts/download.sh, you should have our pretrained model for WikiHop in data/datasets/WikiHop/pretrained-model/. For generating the predictions using this model, follow the steps given in scripts/predict_wikihop.sh.

Training

Follow the steps in scripts/run_full_wikihop.sh and scripts/run_full_obqa.sh for training new models for WikiHop and OBQA, respectively.

Path Extraction Demo

Run the scripts/path_finder_wrapper.py for simply visualizing the paths.

>>> from scripts.path_finder_wrapper import find_paths
>>> documents = ["...this is doc 1 ...", "...this is doc 2 ...", ...]
>>> question = "question text"
>>> candidate = "candidate text"
>>> pathlist = find_paths(documents, question, candidate, style="plain")