Buying Into Semantics: Round 4

The Huffington Post, co-founded by editor-in-chief Arianna Huffington, has turned an investment in semantics, covered here, into an acquisition. The online publisher last year adopted semantic technology dubbed JuLiA from a startup called Adaptive Semantics to help it moderate online postings, which the popular HuffPost says add up to about two million each month.
The purchase is the latest in a recent series of headline-worthy buy-ups on the semantics front, including last week’s purchase of Collexis by Elsevier, Siri by Apple, and of course Radar Networks’ Twine by Evri.
The Huffington Post originally deployed the linguistics algorithms built into the learn-by-example JuLiA system to help take the load off of human editors by automatically deleting the most abusive comments and automatically publishing those that met thresholds for being clean content. The idea was that gives them more time to focus on those comments worth having a legitimate argument about publishing or not.
The always-socially-aggressive Huffington Post isn’t solely relying on tech or in house help to get the job done around identifying appropriate and not-appropriate comments. In addition to their human editors, it’s also introduced various “badge” levels for active contributors to and sharers of its content.
In this beta badge program, a level-one Moderator badge is awarded to those who’ve made at least 20 comments that the HuffPost ends up deleting — and who have a high ratio of good flags to mistaken flags. If you don’t maintain a sufficiently high ratio of successful flags to mistaken flags, you may lose your Moderator Badge, by the way. (It appears to be a way of making sure moderators don’t abuse privileges and flag stuff they may not like but isn’t actually offensive). Recommend 100 comments the HuffPost winds up deleting, and you get trusted with the ability to delete inappropriate comments from the site yourself. Other badges include Networker (levels for those who are most connected and have connected their HuffPost accounts to their Twitter and Facebook accounts), and Superuser, who are even busier sharing and connecting across social networks.
All these protections don’t necessarily mean that some comments tht may have questionable characteristics won’t slip through the tech or human editor cracks. For example, when The Semantic Web blog spoke to Adaptive Semantics last year, co-founder Elena Haliczer noted that it’s very easy to fool key word and spam filter systems by replacing the letters of some obscene words with exclamation marks or ampersands. A recent Superuser post in response to another comment on an article on Michelle Obama sporting glasses while enjoying a sports event reads, complete with ampersands to disguise words:

Not sure where the blame lies there. It is possible the Huffington Post hasn’t yet put the tech to work on all its comments (Adaptive Semantics was handling about 25 percent of the moderation when we spoke back in July). And from both the tech and human perspective, could just be that today bitch isn’t considered an inappropriate word for public consumption, so it would get by with or without ampersands. As for bleeping out the ‘i’ in bigot, perhaps the writer was concerned that falls under ad hominem attack rules. (HP cites as its rules for blocking a comment being extremely abusive, off-topic, using excessive foul language, or including an ad hominem attack.)
Adaptive co-founders Elena Haliczer and Jeff Revesz are said to be joining Huffington Post as project lead, Social News Technology; and director, Social News Technology respectively. The terms of the deal were not disclosed.
• Don’t forget to propose your startup for our Semantic Web Impact Awards. The deadline is Sept. 15.

The 
Eric Franzon
VP Community
Jennifer Zaino
Contributor
Angela Guess Contributor
semanticweb.com Twitter feed loading...