Triple-Scrapers

Contains automated scrapers that collect data in the format of triples (i.e. [head, relation, tail] tuples).

Install

1. Clone the repo.
2. Change to the top-level directory of the repo.
3. Run ./setup_repo.sh to install dependencies and make needed virtual environments. This is run once to setup the repo.

Setup

1. Run ./setup_env.sh to set needed environment variable and source the virtual environment. This is run once each time you begin working with code in the repo.

Scrape AI2Thor

1. Run python thor/thor_scraper.py. Triples extracted from simulated rooms will be contained in pickle files inside the triple-scrapers/thor/rooms/ folder.
2. To scrape the triples into text files instead of a pickle, see comment in the main function of thor_scraper.py inside the thor folder.