An automated pipeline for the discovery of conspiracy and conspiracy theory narrative frameworks: Bridgegate, Pizzagate and storytelling on the web

tl;dr they trained a bot to detect "X action Y" relationships on social media posts, then tried to generalize the behavior of n=2 as an explanation of conspiracy theories.

• Paper does not admit journalists are ever wrong; ex. compares a social media failed conspiracy to one that was reported by the press.

Inspired by the qualitative narrative theory of Greimas, we formulate a graphical generative machine learning model where nodes represent actors/actants, and multi-edges and self-loops among nodes capture context-specific relationships.

Consequently, determining the underlying narrative framework of a conspiracy theory—its cast of characters, the relationships between those characters, the contexts in which those relationships arise, and the previously hidden events the interpretation of which comprise the conspiracy theory’s action—is difficult.

• People rarely tell complete stories.
• Actor-actant relationships strengthen when they are told, so popular figures are more likely to get used again ("stabilize") over time.

Any storytelling event, such as a blog post or a news report, activates a subgraph comprising a selection of actants (nodes) and relationships (edges) from the narrative framework.

Samory and Mitra pointing out that, “conspiracy theories are often collages of many smaller scale theories”

• Actors are combined in to "supernodes" consisting of context dependent relationships of the actor with other ongoings.
• Actors (people, orgs) are nodes and relationships between groups are the edges.

This hyper-edge can also be represented by a set of three pairwise relationships that are coupled: 1) (“Podesta”, used, “the restaurant, Comet Pizza”);

• A "context" is a way to separate the different ways actants may have relationships with one another.