Human-Subject Experiments for Evaluating Explanations

This is the code used for experiments in the following paper:

Explain, Edit, and Understand: Rethinking User Study Design for Evaluating Model Explanations

Siddhant Arora*, Danish Pruthi*, Norman Sadeh, William W. Cohen, Zachary C. Lipton, Graham Neubig

The 36th AAAI Conference on Artificial Intelligence (AAAI-22).

For dependencies, please check the environment.yml file. To create the same conda environment you can run conda env create -f environment.yml (You might have to edit the prefix in the last line in the file.) Also, please look at requirements.txt file in src directory anr run pip install -r requirements.txt to install dependencies regarding user study.

Please refer to README files for our user study.

Repo layout

.
├── datasets                                            # All Datasets
│   ├── hotel reviews                                   # Hotel reviews Dataset
│   │   ├── original version                            # original version of hotel reviews dataset
│   │   ├── additional_downloads                        # Downloaded by authors
├── src                                                 # All Source files
│   ├── edits_highlighted_word_analysis_folder          # Show percentage of edit on top 20% highlighted word
│   ├── global_stats_folder                             # Analysis regarding the usage of global cues
│   ├── human_perf_analysis_folder                      # Outputs metrics of human understanding
│   ├── mixed_effects_analysis_folder                   # Run mixed effects regression model
│   ├── model_training                                  # Train ML model on reviews dataset
│   ├── README.md                                       # Gives instruction for running user study
└── README.md

Note

If you use the code, please consider citing:

@article{arora2022explain,
  title={Explain, Edit, and Understand: Rethinking User Study Design for Evaluating Model Explanations},
  author={Arora, Siddhant and Pruthi, Danish and Sadeh, Norman and Cohen, William W and Lipton, Zachary C and Neubig, Graham},
  booktitle = {Thirty-Six AAAI Conference on Artificial Intelligence.},
  month = {February},
  year = {2022}
}