Junkipedia is a tool created by the Algorithmic Transparency Institute, a project of the National Conference on Citizenship, to provide a collaborative platform for collecting, monitoring, analyzing, and responding to problematic content online.
Weathering this crisis requires high quality, reliably sourced information. The deluge of rumors, misinformation, scams and hate that continue to flood social media have overwhelmed the capacity for any single entity to address alone. Duplication of efforts, blind spots, and the evolving nature of this crisis make it incumbent upon us to centralize and share data to optimize our collective effort.
Junkipedia provides a platform to achieve this goal by enabling institutions large and small, responsible communities, and individuals to pitch in and identify each and every instance of misinformation, so that new problems are spotted and mitigated quickly. This shared resource makes it easier to triage the messages and focus on those which others have not already addressed, and to leverage existing work instead of recreating it.
At its core, Junkipedia is a database -- a repository for the wide range of junk that spreads online. Junkipedia enables organizations and the public to submit content they find across a range of platforms and annotate it to help explain to others why it is problematic. Junkipedia supports multiple interfaces for submitting information, including directly via web-based forms, email, and SMS, directly from social media platforms, and via a programming interface to allow developers to integrate seamlessly with other systems.
Junkipedia helps to automate the analysis and monitoring of problematic content. When information is submitted to the platform, a series of automated processes extract important data related to the content, like geographic location, engagement data, image text, and notable faces in images. Junkipedia automatically creates an archive of the content, which is important given that some problematic content will be taken down by platforms or users themselves.
The system also automatically matches submissions with other instances of problematic content already in the database. It continually updates engagement data to identify changes that would indicate a shift in its potential for virality and, thus, impact. This gives Junkipedia users insight into the kinds of messages that are spreading (political and cultural themes, the volume of content creation, and engagement) and informs effective inoculation messaging to blunt the force of problematic content.
All of this information is saved as a single entry in Junkipedia called an issue. This granular information in issues can be used to search, filter, classify, and associate different issues with one another. Users can also subscribe to receive hourly, daily, or weekly alert emails to keep them up-to-date on new issues that match their specific focus.
Users on the Junkipedia platform can leverage these features to better understand and critically analyze instances and collections of problematic content in order to identify patterns of behavior, persistent themes, and targeted narratives. In addition, users can classify content according to the threat it represents to the public (high / medium / low), and provide supplemental context and commentary. Issues can be annotated with fact-checks and response messaging that can be automatically sent as responses when users submit new content that matches issues in the database, providing a real-time mechanism for getting high quality information into the hands of the public.
Junkipedia is not a solution to the problem of misinformation. Junkipedia empowers a range of researchers, journalists, grassroots activists, and other institutions to collectively identify and document the problems they find, and add valuable context and advice so that the community can benefit from a centralized shared repository of misinformation