Google will launch ‘revised process’ for improving search engine errors in early 2018

Spread the love

Google will launch a ‘revised process’ for improving search engine errors early next year. According to a spokesperson, it is now complex for users to point out errors to the company in the Knowledge Graph.

The Knowledge Graph is visible in the cards that users receive for certain searches about, for example, a person or organization. Digital assistants such as Amazon Alexa and Google Assistant also rely on this information for spoken search results in smart speakers, for example. Correcting a bug in the Knowledge Graph currently takes between three weeks and three months, The New York Times reports.

It’s unknown what the revised process will look like, but it should provide better support if automated systems don’t work well enough. The company does not have a phone number for support, relying almost exclusively on web pages and computer systems. A Google spokesperson told the paper that it’s important to keep a balance between quick changes and correct changes. It is unknown how many people turn to the search giant every year to correct errors.

The newspaper asked Google about the process after a newspaper employee spent a week trying unsuccessfully to get incorrect information from the Knowledge Graph. Google provided information that she had passed away four years ago. That information was about someone else with the same name. Her photo was taken from the site of The New York Times.

This is not the first time that the Knowledge Graph has displayed incorrect information. In March it turned out that the algorithms that Google uses are difficult to separate fact from fiction. For example, based on the Wonderopolis site, Google answered the question of why fire trucks are red with a statement from the British comedy series Monty Python. In other cases, Google has claimed that former United States President Obama is planning a coup and that certain politicians are members of the Ku Klux Klan.

You might also like