Product Case-Study: Live Tracker

Product Case-Study: Social Alert

Aggregation of human analysis

News organizations, expert collectives, and user groups are identifying and flagging misinformation.

We aggregate their analysis through open/paid APIs and check their validity by cross-referencing datasets and sources.

Natural Language Detection (NLD)

By analyzing large datasets, misinformation can be detected in both written and spoken word.

Sentiment analysis is the task of extracting emotions and stance detection is the task of assessing what side of a debate an author is on from the text. Fact checking compares extractable facts with credible databanks.

Our pillars for validating information

Network/Source Analysis

Most widely-used method to uncover misinformation.

Based on identifying patterns in the network of all accounts talking about a topic.

 

Shared connection, suspicious behavior, and distributed content allow us to draw conclusions about activity from propaganda networks.

Time Series Analysis

This method is based on evaluating characteristics of non-real and real news in terms of volume and velocity with reference to time series.

By following sources of content pieces over time, the "way" they spread over platforms allows statistical relevant determination of their credibility.

  • Instagram
  • White Twitter Icon
  • White Facebook Icon
  • LinkedIn