Wikipedia Develops Application with AI

The Wikimedia Foundation just released a new service engineered to heighten the accuracy and detail of Wikipedia articles.

Known as the Objective Revision Evaluation Service (ORES), the application uses artificial intelligence (AI) and machine learning (ML) to help Wikipedia editors to identify bad articles with faster speeds and assign them with appropriate scores almost immediately.

Unknown to most Wikipedia users, Wikipedia is actively edited over 500,000 times a day. Volunteer editors must then review those changes, making a lot of man power necessary to keep the site alive and accurate.

oresORES makes it easier for these editors to look through and organize incoming content, identify poor edits and mark them for further scrutiny. Bad edits occur often and range from the deletion of accurate information, addition of inaccurate information, addition of an opinion and addition of an obscenity.

Principal analyst of the Enderle Group Rob Enderle had this to say: “If you’re in the media at all, there’s a chance that someone is going to dislike something that you said and is going to try to damage your Wikipedia page.”

“Low-level AI is really good at identifying patterns and taking prescribed action against the patterns it recognizes. Unless you have a ton more people than Wikipedia has, you’d never be able to keep up with the bad edits. Wikipedia can be more trusted and less likely to be used as a tool to harm somebody [now that it has developed ORES]).”

Wikimedia Senior Research Scientist Aaron Halfaker reiterated the improvement in accuracy and timely edits that ORES is sure to bring: “That allows the editor to review the most likely to be damaging edits first. That can reduce the workload of reviewing edits by about 90 percent.”

So how does it work? ORES is engineered in such a way that it can predict the probability that an edit is damaging by checking in on the before-and-after edits common to all articles that appear on Wikipedia. It then assigns a score to a proposed edit depending on its likelihood of being damaging.

half“Our machine learning model is good enough at sorting those edits by the probability that they’re damaging that you would have to review 10 percent of the incoming edits to know that you caught all of the damaging edits,” continued Halfaker. “Without this tool, you’d have to review all the edits to know you caught all the damaging edits.”

“One of the reasons we want to reduce the workload around quality control is so that editors can spend more of their time working on new article content rather than removing vandalism.”

Despite Wikipedia’s reception of about 12 million hours of volunteer labor a year, they could never get enough volunteers to manage all of the internet vandalism that occurs under their business model in a timely and reliable manner.

Leave a Reply

Your email address will not be published. Required fields are marked *