Cite-Unseen

Cite Unseen is a JavaScript script for the English Wikipedia that automatically adds iconic indicators to citations to quickly convey the nature and bias of sources to readers.

This project was first developed by Kevin Payavi and Josh Lim. The project was started at the CredCon 2018 hackathon in Austin, Texas, and has received support from the Credibility Coalition and Knowledge Graph Working Group.

Links

• Website and full documentation: https://en.wikipedia.org/wiki/User:SuperHamster/CiteUnseen
• Slidedeck: https://commons.wikimedia.org/wiki/File:Cite_Unseen.pdf
• Phabricator Project (issue tracking): https://phabricator.wikimedia.org/project/view/4050/

Repository Overview

• index.js - the script that runs on each Wikipedia page load. The script iterates over each citation on a Wikipedia article, pulls its external links, and compares them to the categorized domain data. The categorizedDomains and categorizedStrings objects need to be populated before running.
• /scripts - various scripts used to scrape and generate data for Cite Unseen. Most of these scripts are rather crude with room for improvement.
• /data - data files (see below for more)

Data Overview

• data/categorized-domains.json - partial and full domains that are categorized by type (government, opinion, community news, etc.). In the script, each entry is prepended with a . and // before doing comparisons (e.g. an entry for wikipedia.org will be compared as .wikipedia.org and //wikipedia.org in the script, to account for cases like en.wikipedia.org and https://wikipedia.org).
• data/categorized-strings.json - other strings that might appear as part of a domain (such as /opinion/ for finding opinion sources).