We share the code of our ICWSM 2020 paper : Towards Quantifying the Distance between Opinions.

1. First, create a fresh virtual environment and install the requirements.

    conda create -n opinion_distance_env python=3.6
    conda activate opinion_distance_env
    conda env create -f environment.yml
   

2. Get the TagME API gcude-token from here and add the obtained gcude-token in the file present at files/tagme_gcude_token.txt.

3. Download word2vec embeddings from here and store the file GoogleNews-vectors-negative300.bin.gz in embedding folder.

You can run Opinion distance on Seanad Abolition Dataset as follows

python src/run_opinion_measure.py --path dataset/CiviQ_Seanad/ --embedding_strategy word2vec --semantic_threshold 0.6 --baselines False

The above code will generate

1. Clustering result in folder dataset/CiviQ_Seanad/results
2. Compute opinion distance matrices in folderdataset/CiviQ_Seanad/dist_mats

Note that the shared code computes opinion distance using OD method mentioned in the paper and uses word2vec embedding strategy. The code requires Internet connection to compute opinion representation using TagME API.

The classification of opinion can be performed by running the below code

python src/supervised_opinion.py dataset/CiviQ_Seanad/ 

If you find our paper useful in your research, we ask that you cite the following paper:

Gurukar, Saket, Deepak Ajwani, Sourav Dutta, Juho Lauri, Srinivasan Parthasarathy, and Alessandra Sala. "Towards Quantifying the Distance between Opinions." ICWSM 2020.

@article{gurukar2020towards,
title={Towards Quantifying the Distance between Opinions},
author={Gurukar, Saket and Ajwani, Deepak and Dutta, Sourav and Lauri, Juho and Parthasarathy, Srinivasan and Sala, Alessandra},
journal={International Conference on Web and Social Media (ICWSM)},
year={2020}
}

For questions or comments about the implementation, please contact gurukar.1@osu.edu and deepak.ajwani@ucd.ie.