Acting as a neutral, “White Label” the NCRI seeks to innovate exacting internal standards of objectivity, data security, and strict legal compliance. It seeks to earn the trust of the public and social networks and merit the trust it earns in order to afford transparency.
- We collect billions of posts from fringe and mainstream websites and millions of images which we analyze with artificial intelligence.
- We issue peer-reviewed reports and data-journalism on our blog that detail our findings to the public to help us better understand and hopefully treat the ominous infection of hate in modern political life.
Some examples of the racist and anti-semitic “happy merchant” meme
Visualization of a subset of the obtained image clusters with a particular focus on the penetration of the Happy Merchant meme to other seemingly neutral memes.
Graph representation of the words associated with “jew” on the alt-right-affilicated messaging board /pol/. We extract the graph by finding the most similar words (cut off at 0.4 cosine distance value), and then we take the 2-hop ego network around “jew.”
Graph representation of the words associated with “white” on the alt-right-affilicated messaging board /pol/. We generate the graph using the same procedure as Fig. 5.
A 2-hop neighborhood/visualization of how the disgusting slur, “ni****” is used on the alt-right-affilicated messaging board /pol/. Modeleted from millions of comments, distance equals closeness in relatedness, while size indicates frequency of term use.