A repository for hosting code and dataset, and managing the workflow for the paper.
This repository contains code to extract privacy policies, create a dataset, create and save the word embeddings from the dataset. Specifically it contains code to work on the 1,010 privacy policies from Fei Liu et. al's work.
Liu, Fei, Nicole Lee Fella, and Kexin Liao. "Modeling language vagueness in privacy policies using deep neural networks." 2016 AAAI Fall Symposium Series. 2016.
The dataset can be downloaded from the Usable Privacy Org Website. To download click on the ACL/COLING 2014 Dataset.
There are two modules that can be useful when using the 1,010 privacy policies - create_dataset.py and create_embedding.py. The create_dataset.py module extracts all the data for each privacy policy and saves it as a json file.
The module uses commandline arguments to create and save the dataset. Following is an example for how to use -
$ python src/create_dataset.py --dataset_path <enter-path-to-dataset-folder>/ --json_file <enter-filepath-with-name-of-json-file>
$ python src/create_dataset.py --dataset_path code/acl_dataset --json_file code/acl_dataset/dataset_as.json
Another module is the create_embedding.py module which takes in the path to the dataset and filepath to save the embedded words. The module uses Gensim algorithm to create the embeddings.
$ python src/create_embedding.py --dataset_path <enter-path-to-dataset-folder>/ --dim <enter-dimension-size-of-embedding> --embedding_file_path <enter-filepath-with-model-file>
$ python src/create_embedding.py --dataset_path code/acl_dataset/ --dim 100 --embedding_file_path code/acl_dataset/embedding_dim_100.model