LONDON: Google has developed technology that will change search engine results from showing the most popular web page to the most factually accurate contents on the page.
research paper published by Google in February shed some light on how this problem could be fixed. Their solution was to rank search results by the accuracy of the facts and not by popularity.
What Google deems “facts” are determined by Google’s “Knowledge Vault,” which the researchers described as “a database of 2.8 [billion] facts extracted from the web.” Facts included in the vault are based on “knowledge triplets,” or corollary pieces of information selected by Google and sought out for confirmation online.
Sometimes fake facts on the internet are harmless fun – entertainment that is liked and shared simply because it’s entertaining. However, there are instances where websites climb the rankings that shouldn’t. This anti-vaccination website is one of the top search results for “vaccination,” for instance, even though it is full of information that is either wrong or harmful to children. (And the fact that Business Insider just linked to it has only compounded its superior ranking within Google’s results.)
At its core, Google ranks web pages based on the number of incoming links they receive. The assumption is that the more links a page has, the more important it must be on the web. The algorithm has been adjusted and modified hundreds or thousands of times over the years, of course, but incoming links are still a huge part of what determines any site’s ranking in a search. Google’s engineers adjust the algorithm periodically in hopes of making sure it returns the highest quality searches, not simply the most popular sites.
To weed out popular lies, Google has devised a method/model that measures the “truthfulness” of a web page instead of its online reach. A post on a blog might have a big reputation, but that doesn’t always mean it’s factual. As NS explains, instead of counting incoming links (a measure of its reach) Google’s new system could count the number of “facts” in the page. Each source is then analysed for how many lies it has and scored on that using something called a “Knowledge-Based Trust” score.
Google used its “Knowledge Vault” to qualify the information. That’s the company’s giant database of information, vetted facts and research.
Google’s lie detector isn’t live yet and it’s unlikely to launch any time soon. At this stage it’s simply a research paper published by Cornell University called “Knowledge-Based Trust: Estimating the Trustworthiness of Web Sources.”
