Google is leveraging BERT which is one of the AI language models that can provide full news stories coverage for matching the better stories with facts checks and also to have a better understanding what are the most relevant results to the queries that are being posted on the search. The Google BERT is a natural language processing (NLP) framework and BERT has altered significantly the NLP landscape and has been inspired by many of the recent NLP architectures for training approaches and also language models.
BERT is a most advanced AI-based system which has capabilities to understand more complex, natural language queries. But when it comes to high-quality information which is trustworthy and has advanced information understanding capabilities because google doesn’t understand the content the way which the human can understand.
But this search engine can largely understand the quality of the content through by the way knows as signals. Danny Sullivan who is the Public Liasion of Google says that the number of the quality of the pages that get linked to the particular page is the signal of that page which can be a trusted source about the topic. Google has 10,000 more than search quality raters who collectively do millions of the sample searches and also rate the quality of the result.
The company had also helped to make it easy by spotting the fact checks in the searches, news, and also google images by providing the display fact check labels. These facts check labels are generally provided by the publishers who mostly use the ClaimReview schema for marking the facts checks that are mostly getting published. But Sullivan also said that the Google system is not always perfect. Also if the system gets failed to protect the privacy policy violating the appearing content, the Google enforcement team will take action according to the policies.
Google is also closely working with Wikipedia for detecting and removing vandalism which may be used in the knowledge panels. The search engine giant now can detect the breaking news queries within a short span of minutes rather than 40 minutes which it took earlier.
