Home Stamp collecting ADL Develops New Algorithm to Track Antisemitism Online | IJN

ADL Develops New Algorithm to Track Antisemitism Online | IJN


By Asaf Shalev

NOTEW YORK — When it comes to anti-Semitism on social media, the algorithms governing the major platforms are partly responsible for their reach.

ADL CEO Jonathan Greenblatt pictured in 2018. (Michael Brochstein via Getty)

The ADL announced March 8 that it had built an algorithm called the Online Hate Index, describing it as the first tool ever developed to measure anti-Semitism on social media platforms.

The program can quickly sift through millions of posts to detect anti-Semitic comments and assist in their removal.

This system uses an algorithm informed by artificial intelligence to find and classify messages as potentially anti-Semitic. These messages are then passed on to a team of volunteers and experts, who use their judgment to make the final decision.

The system also checks whether the messages are ultimately deleted.

The new index was needed because social media companies aren’t transparent enough about their efforts to curb the spread of hate speech on their platforms, ADL CEO Jonathan Greenblatt said.

“We will use this tool to hold social media platforms accountable for how they proactively remove hate and how their content moderators respond to reports,” Greenblatt said.

One of the objectives of the project is to demonstrate that if the LDA can developing the technology to track anti-Semitism, surely Silicon Valley can do that too.

Social media companies have tried to tackle anti-Semitism in the past. Facebook (now known as Meta) stumbled over its decision to ban Holocaust denial on its platforms; engineers developed screens that also sometimes blocked legitimate educational messages intended to raise awareness of the Holocaust.

FIn its first analysis, the ADL used its system to scan Reddit and Twitter, collecting posts for a week in August last year.

The ADL chose these platforms because they are the only major ones to offer open access to their data. Facebook, on the other hand, generally does not allow outside groups to access searches.

The algorithm used by the ADL was trained to spot instances of possible anti-Semitism. In a process known as machine learning, humans labeled comments as anti-Semitic and fed them to the algorithm, which in turn began to recognize patterns. The more comments the algorithm processed, the more effective it became at catching anti-Semites.

Anti-Semitic statements such as “Jews are lizards prove me wrong” and “The Magic of Jewish Mind Control” were among the roughly 2,000 Reddit posts identified by the ADL system, out of some 40 million total comments added to Reddit during this week.

The number of people who view a comment on Reddit is partly determined by whether users “vote up” or “vote down” — and there’s good news in that regard. Users get on average a third fewer anti-Semitic comments than other types of posts, according to a report published by ADL about its analysis.

“Statistical analysis of these scores shows that anti-Semitic content on Reddit is significantly less rewarded than non-anti-Semitic content,” the report said.

For Twitter, ADL estimated that there were some 27,400 anti-Semitic tweets among the 440 million posted during the week its software was reviewed, and that those tweets could have been seen by up to 130 million people.

The ADL warned that it designed its net conservatively and only looked at English text, meaning video, audio and images were excluded, along with anything written. in a foreign language.

On both platforms, most anti-Semitic comments remained posted for months after being posted and were not removed even after the ADL alerted the platforms.

OOne of the challenges in any attempt to eradicate anti-Semitic discourse is to define the term. A The contentious issue is deciding when criticism of Israel crosses the line into anti-Semitism.

The ADL report says its algorithm is trained by in-house experts and volunteers from the Jewish community.

This does not mean that human judgment is entirely outsourced to computers. In the ADL’s system, artificial intelligence is used to sift through masses of content, with its human teams ultimately determining which posts constitute anti-Semitism.

To help them with their decisions, each volunteer receives an introduction, which includes a reference to the definition of anti-Semitism drafted by the International Holocaust Remembrance Alliance, which has proven controversial because it focuses on anti-Israel discourse.

Some examples in the alphabet include “claiming that the existence of a State of Israel is a racist enterprise” and “denying the Jewish people their right to self-determination”.