System to detect, assess and counter disinformation
A computer-readable medium for the identification, measurement, and combatting of the influence of large-scale creation and distribution of disinformation is herein disclosed. An embodiment of this invention is comprised of one or more repositories of data which involve online comments and articles and attributes derived from them, one or more technical targeting systems, a content analysis system, a cost and influence estimation system, a dialog system, a performance management system, a bot design and test system, a security system, a multimedia content generator, one or more machine learning components, a data collection mechanism, separate consumer and human operator applications, and a mechanism for the creation and management of bots across multiple channels.
1. A technical targeting system for multi-evidence targeting, comprising:
graph processing server configured to accept messages from one or more message queues of online content, the one or more message queues fed by a data collection process encoding online content into data records for the one or more message queues,
one or more storage devices configured to store data and manage a knowledge base; and
one or more processors, configured to execute functionality for subsystems, components, and mechanisms stored in one or more non-transitory computer readable mediums, the one or more non-transitory computer readable mediums store subsystems, components, and mechanisms comprising:
a Natural Language Understanding/Natural Language Generation (NLU/NLG) subsystem configured to reference the knowledge base;
an assertion and event identification, extraction and analysis component configured to execute the NLU/NLG subsystem to extract the data from the accepted messages for storage into the one or more storage devices, the extracted data directed to assertions or references to real world events,
an influence measurement component configured to detect sentiment change from the stored data and label disinformation in the stored data based on the detection of coordination patterns,
a narrative and campaign identification component configured to identify one or more narratives from one or more sequences of disinformation from the stored data and one or more campaigns of disinformation regarding the real-world events within a bounded window of time.
2. The technical targeting system of claim 1 , wherein the one or more processors are configured to:
identify an at least one characteristic of disinformation based on one or more of:
presence of following processes, commands, and work schedules,
coordinated behavior,
sock puppet transfers,
anomalous echo chamber formulation,
content transmission patterns,
aspects of the online content being posted,
trajectories,
accounts being transitioned among multiple identities,
narrative transmission,
campaign participation,
inauthenticity in one or more of demographic, professional or domain
knowledge, employment status, physical location, and bot versus human,
characteristics displayed by bots,
behavior when confronted with system bots,
unstable lexical fingerprints,
malicious or adversarial actions,
unusual regularities in behavior, or
correlation to unusual changes in user behavior, Social Network Analysis (SNA), and image component reuse.