A analysis collaboration between Vahid Ghafouri and Guillermo Suárez de Tangil from IMDEA Networks Institute, together with Waleed Iqbal, Gareth Tyson, and Ignacio Castro from Queen Mary College of London, explores the methods during which real-world inequalities are mirrored in social networks.

Their examine, titled “Woman and the Tramp Nextdoor: On-line Manifestations of Actual-World Inequalities within the Nextdoor Social Community,” was offered on the ICWSM 2023 convention. The analysis reveals that an individual’s earnings stage influences the kind of content material they share on-line.

The researchers analyzed 2.6 million posts from 64,283 areas within the U.S. and three,325 neighborhoods within the U.Ok. printed on NextDoor, a social community the place customers focus on issues of their neighborhoods. By analyzing the places connected to every submit, the researchers estimated the customers’ earnings based mostly on the typical earnings of these places supplied by the governments.

The evaluation confirmed that people residing in wealthier neighborhoods are inclined to take part in discussions about crime extra regularly than these in poorer neighborhoods. This commentary means that there could also be a necessity for policymakers to pay extra consideration to crime points and allocate assets accordingly.

Vahid Ghafouri, a Ph.D. scholar at IMDEA Networks and UC3M and co-author of the examine, acknowledged, “Surprisingly, this sample persists although actual crime charges are decrease in wealthier neighborhoods. This means that wealthy persons are extra delicate and reactive to the presence of crime.”

The analysis additionally revealed that nonviolent crimes had been mentioned greater than violent ones by each Individuals and British customers. Nonetheless, rich Individuals centered extra on violent crime and weapons in comparison with their counterparts within the U.Ok.

Moreover, the examine discovered that residents of wealthier neighborhoods expressed extra constructive sentiments of their on-line texts, indicating a possible correlation between earnings and happiness. This pattern was constant throughout each the U.S. and the U.Ok.

This groundbreaking analysis showcases the potential of AI and Pure Language Processing in predicting earnings, inequality, and crime charges based mostly on user-generated on-line content material. It has been printed within the Proceedings of the Worldwide AAAI Convention on Internet and Social Media.