Tel Aviv start-up helps websites tone down hateful comments

Spot.IM has developed ‘fully automated tools’ to monitor violent language.

By MICHAEL ZEFF
Updated: DECEMBER 28, 2016 03:03
Cyber hackers [illustrative] (photo credit: REUTERS)
Cyber hackers [illustrative]
(photo credit: REUTERS)
In recent years, the comment sections in online media outlets have become synonymous with hate and verbal violence due to the perceived anonymity of the commenters.
Israeli start-up Spot.IM developed a platform for websites that seeks to regain control over online discourse and tone down the violence.
A study conducted on behalf of Tel Aviv’s Spot.IM was published earlier this week, shedding light on the increase in the levels of verbal violence and hateful online discourse, especially surrounding online news publishers.
“Our general goal is to build and maintain online communities for publishers. Especially today, when most of the discourse is done through social networks, we see the negative effect it has on the way people consume and react to news content. We are trying to put these conversations back with the publishers, and part of that is indeed managing comment sections and monitoring them,” Roy Laor, Spot.IM’s head of partnerships, told The Jerusalem Post.
Founded in Israel in 2012, Spot.IM is now headquartered in New York. According to their site, the company works with more than 25% of the US-based digital publishing industry, including The Huffington Post, AOL, Time and Forbes. The company’s software registered traffic from more than 410 million unique devices all over the world that amount to an average of 40 billion views a day.
“Recently, we noted a growing global demand to reign in online violence and prevent offensive content from being published. On social media, users are required to identify themselves, but when working under a shroud of anonymity [in comment sections] they allow themselves to be more violent and need to be filtered,” Nadav Shoval, Spot.IM’s founder and CEO, said.
Laor, who manages the company’s partnerships with clients like Time and Forbes, explained that the increase in violent language and trolling in the comments section kills debates on news sites, moves them away from the publisher and damages the publisher’s reputation.
“On the other hand, publishers don’t really like the task of monitoring the comments sections. First, it just isn’t fun, all day long one is exposed to the horrible things that people write online. Second, it’s a very tiring task. So we developed fully automated tools that do this with minimal human involvement on the publisher’s side,” Laor told the Post.
The company’s platform, which is integrated into the Post’s online edition, is also found in more than 5,000 active news sites and online magazines in the US, Israel and all over Europe.
According to the research conducted by Spot.IM, Israel leads the way with the highest relative amount of comments deleted due to their content, 6.5%. Russian commenters take second place with 5.5% of all comments deleted, the US follows with 5% and finally, all of Europe combined, with 4%. This data constitutes an overall increase in violent language all over the world.
But Laor explained that these numbers are not absolute. In Israel, for example, even nonviolent posts can be taken down due to the IDF censorship on articles dealing with security issues. In the US, it is significantly more difficult to delete every single hateful comment, even if the publisher really wishes to do so, due to the First Amendment of the Constitution.
“Additionally, we’ve noticed that the online discourse in Israel grew more violent and offensive in the last year. In the US, we noted a similar increase in and around the elections.
This means that publishers have also been deleting and blocking more comments than before,” Laor said.
According to Spot.IM, their comments management tool can identify individuals who prowl the Internet hoping to spark violent discourse or aggravate them. The program can then help filter their activity up to and including blocking them and deleting their entire comment history. Even if trolls believe themselves to be anonymous and move from one site to another – as long as both sites operate with Spot.IM – the company’s platform can identify them, follow their moves and take action against them.
“Our platform can identify an anonymous user through various means, such as an IP address and cookies. But even if users bounce their IPs by using various proxies, we can still identify them and track them across sites. If trolls get blocked by the publisher of one site and move on to another, we can then flag their first post with the publisher of the new site, and that site can choose to take action as well,” Laor explained.
This is done through the platform’s “machine-learning” artificial intelligence, which can study and retain recurring patterns of violent, inappropriate and hateful comments – not only through obvious keywords but also through sentence structures and recurring spelling and grammar mistakes.
According to Laor, one particular example in the US occurred during the presidential elections.
An anonymous commentator who used extremely violent language that was unacceptable even taking into account the strict standards of freedom of speech would always use uppercase letters in a recurring fashion. Spot.IM identified the pattern and began automatically informing the publisher each time the anonymous poster made a new comment.
“Our goal is to clean up these phenomena from online communities in partnership with publishers by giving them the tools to deal with it and maintain their reputation,” Laor said.


Related Tags
terrorism on the internet