The Open Event Data Alliance Software Page
This is the Github landing page for the software being developed by the Open Event Data Alliance. For more information about OEDA, please visit the homepage. In a nutshell, the goal of the OEDA is to enable the generation and sharing of open, replicable event data generated from news sources. We have developed a series of software tools to generate event data, which are described below. Information on event data more generally can be found on the Computational Event Data System website.
PETRARCH is the Python-language successor to the TABARI event data coding software. PETRARCH makes use of a full parse, based on Stanford CoreNLP, to better perform word disambiguation and noun and verb phrase chunking. This full parse allows for more accurate event coding. PETRARCH is currently under active development, and we aim for a full release in the summer of 2014.
More information can be found at https://github.com/openeventdata/PETRARCH-Development with documentation at http://petrarch.readthedocs.org/en/latest/.
Event/Location: Dataset In A Box, Linux-Option or event data in a box, basically. EL:DIABLO is a Vagrant box that enables the easy creation of a virtual machine on the end-user’s machine containing all the tools needed for generating event data (including the scraper and pipeline described below). Our hope is that this box will allow others to easily replicate the system we use to generate event data. This will allow others to examine, critique, and improve upon our system.
We make use of a web scraper with a whitelist of RSS feeds to pull news stories from ~160 unique websites. These scraped stories are stored in a MongoDB database for easy future retrieval.
The pipeline is what moves from the web scraper, through the event coder, and finally to coded event data. In short, the pipeline is the glue that holds all of the various components together.
More information can be found at https://github.com/openeventdata/phoenix_pipeline with documentation at http://phoenix-pipeline.readthedocs.org/en/latest/.
Work by Philip Schrodt was partially supported by the National Science Foundation Political Science and Methods, Measurements and Statistics programs, Grant SES-1259190.
Work by John Beieler and Muhammed Idris was partially supported by National Science Foundation under IGERT Award #DGE-1144860, Big Data Social Science, and Pennsylvania State University.
Much of this work has been, and continues to be, supported by Caerus Associates in one form or another.