Google Unveils Improved “Bert” Algorithm
In a media briefing on Thursday, Google provided insights into its new search algorithm known as “Bert,” or “Bidirectional Encoder Representations from Transformers,” which is one of the company’s largest improvements to its search algorithm in the last half decade. Purportedly, Google’s new algorithm will allow 1 in 10 search queries to return better results. The company has written Bert such that it is able to better understand the multiple meaning of words and slang to produce more relevant results. For example, for the query “do estheticians stand a lot at work,” the current algorithm produces results about free-standing beauty schools, while Bert is able to parse the user’s meaning and return a result on the physical demands of estheticians. These changes are slowly being rolled out, and users are will soon be able to feel the effects of this search algorithm.
The article does not give too much detail on how Google’s advertising is being affected by this new algorithm, with a spokesperson remarking that Google is simply exploring how Bert can be useful for their various products. I strongly suspect that while Bert may help to improve the results of people’s searches, it may also be a vehicle to disproportionately boost the activity of not just Google products, but also Google sponsored pages. On the other hand, it is indeed exciting to see that Google is trying to improve the user experience. This article is highly relevant to the class because Bert may eventually be involved in advertising on Google or sponsored pages, and may affect the way the matching market for advertising is run on Google. Furthermore, this article provides insight on how search engines are able to traverse the information on the web, which provides a real world example of the interconnectedness of the web and the networks that are formed through hyperlinks.