Human Motive Identification

This repository contains code and data for reproducing experiments in our paper ``Toward Comprehensive Understanding of a Sentiment Based on Human Motives.''

Setup

Data

For setting up data, see data/README.md.

In addition, download GloVe embeddings and place it at embeddings/GloVe/glove.6B.100d.txt.gz.

Code

You need to install allennlp (for an MLP classifier) and sckit-learn (for an SVM classifier) to run classifiers.

pip install allennlp
pip install scikit-learn

Run experiments

SVM-BoNG

PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/linsvm_1_res.json -v
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/linsvm_1_lap.json -v

SVM-BoNG-tfidf

PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/linsvm_2_res.json -v
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/linsvm_2_lap.json -v

MLP

# SWEM
PYTHONPATH=. python classifiers/allennlp/run_mlp_cv.py config_files/mlp_swem_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_mlp_cv.py config_files/mlp_swem_1_lap.json -s <result dir> --include-package classifiers.allennlp

# CNN
PYTHONPATH=. python classifiers/allennlp/run_cnn_cv.py config_files/cnn_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_cnn_cv.py config_files/cnn_1_lap.json -s <result dir> --include-package classifiers.allennlp

# BiLSTM
PYTHONPATH=. python classifiers/allennlp/run_lstm_cv.py config_files/lstm_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_lstm_cv.py config_files/lstm_1_lap.json -s <result dir> --include-package classifiers.allennlp

Transfer learning

# SVM-BoNG
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/transfer/linsvm_1_res.json -v
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/transfer/linsvm_1_lap.json -v

# SVM-BoNG-tfidf
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/transfer/linsvm_2_res.json -v
PYTHONPATH=. python classifiers/sklearn/run_cv.py config_files/transfer/linsvm_2_lap.json -v

# SWEM
PYTHONPATH=. python classifiers/allennlp/run_mlp_cv.py config_files/transfer/mlp_swem_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_mlp_cv.py config_files/transfer/mlp_swem_1_lap.json -s <result dir> --include-package classifiers.allennlp

# CNN
PYTHONPATH=. python classifiers/allennlp/run_cnn_cv.py config_files/transfer/cnn_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_cnn_cv.py config_files/transfer/cnn_1_lap.json -s <result dir> --include-package classifiers.allennlp

# BiLSTM
PYTHONPATH=. python classifiers/allennlp/run_lstm_cv.py config_files/transfer/lstm_1_res.json -s <result dir> --include-package classifiers.allennlp
PYTHONPATH=. python classifiers/allennlp/run_lstm_cv.py config_files/transfer/lstm_1_lap.json -s <result dir> --include-package classifiers.allennlp

Citation

Naoki Otani and Eduard Hovy. 2019. Toward Comprehensive Understanding of a Sentiment Based on Human Motives. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL), July. Association for Computational Linguistics.