Employing the 'Wisdom of Crowds' to Warnings About Unsafe Toys

Dec 21, 2015

Credit https://creativecommons.org/licenses/by/2.0/

With all the gift-giving this time of year, it’s important to know you’re not giving a child a toy that might be dangerous.  Scientists at Virginia Tech have come up with a new way to flag potential toy hazards before they cause injuries. 

These days there’s no shortage of product reviews out there.  In fact there are too many for consumers to actually comb through. Now a technique for mining those product reviews, invented at Virginia Tech’s Pamplin School of Business to examine car safety, is now being applied to children’s toys.

Matthew Winkler is a senior majoring in business and information technology. He’s the lead author on a soon to be published study on how certain words in a review, the researchers call smoke words signal early warnings of hazards even if none has occurred. And that’s significant, because currently the Consumer Product Safety Commission responds to toy hazards only after they cause injury.

“Using the smoke word method we can, before the CPSC tests it or responds to a call of an injury from a parent, we can use an automated system to prevent a lot of that from occurring and hopefully save a couple of lives even.”

Associate Professor of Business and Information Technology, Alan Abrahams is part of the team that came up with the text-mining program. He says it’s not just about checking the one star reviews for red flags because sometimes, clues to hazards are often embedded in positive reviews.  They may like the product, and it may not have caused injury, but people often extrapolate potential problems that they see could happen and then warn others about them in their reviews.

“The ability to make use of the wisdom of the crowd of consumers intelligence about what might be a concern with a toy, I think can give the Consumer Product Safety Commission a workforce of literally thousands of concerned parents as opposed to merely a few hundred product specialists.

Once the study on text mining for toy safety is published, its creators hope the new technique will be used by toy manufacturers, retailers, consumer safety agencies and consumers, to prevent injuries.  Currently, according to Center for Injury Research and Policy, toy-related incidents send a child to the emergency room somewhere in this country, every three minutes.