Artificial intelligence technology "RankBrain" with the potential to drastically change the precision of Google search

Google develops artificial intelligence "Deep MindIt is known that it acquired the artificial intelligence and machine learning technology, and attention was focused on how to utilize and apply this artificial intelligence and machine learning technology. A Google engineer uses the artificial intelligence technology applied technology in Google search engine "RankBrain"Is already in operation and reveals that it exerts more effect than imagination in order to display more optimum search results.

Google Turning Its Lucrative Web Search Over to AI Machines - Bloomberg Business
http://www.bloomberg.com/news/articles/2015-10-26/google-turning-its-lucrative-web-search-over-to-ai-machines

In Google's search system engineer Greg Collard, in an interview with Bloomberg, the Google search engine revealed that it has introduced a new signal "RankBrain" that applied artificial intelligence for several months.

The state of the interview with Mr. Colade is confirmed by the movie (note that the sound is played automatically) on the site above.


According to Mr. Colado, in a situation where key phrases and keywords etc. are not clearly known, in a situation where we conduct "vague search" which searches for something by combining fragmentary information, RankBrain combines information While searching for information close to the contents, past search contents andQueryWhile referring to, you can display what you want. It seems to be superior also in terms of being able to infer the optimum answer by interpreting the contents from the surrounding information even for queries which have not been seen so far. In addition, Google search has generated millions of queries per second, of which 15% are unknown queries, Collard reveals that RankBrain is processing most of these queries.


In addition, RankBrain differs from the existing "signal", which influences the search rankings that Google engineers built, in terms of "being able to learn," Collard said. This means that RankBrain has the ability to improve its search accuracy by brushing himself up by learning the phrases searched and the query itself. Beyond this technology, the ability to understand users' behavior, such as what pages they browse and what kind of links they followed, to the retrieval results displayed, so that the retrieval accuracy can be increased more and more It can easily be imagined that it will be given.

According to Corrado, the accuracy of RankBrain has already exceeded that of Google engineers. In exams showing some pages and hitting the page that comes to the top of the search, Google Engineer's correct answer rate was 70%, while RankBrain was 80%.

Under the circumstances where search trends are changing due to the popularization of mobile terminals, Google considers that the accuracy of search using surrounding circumstances such as ambiguous searching and searching by disjoint phrases is an important point that affects quality of service In this regard, RankBrain, which can infer the unknown information by combining various elements and refine himself by learning himself seems to be a powerful weapon. According to Corrado, among the hundreds of signals embedded in the Google search engine, RankBrain has already just started to top three just a few months after its appearance.


From Mr. Collard 's remarks, RankBrain is likely to be considered a tool for processing queries rather than signals that affect search ranking. Either way, by developing a technology that enhances search accuracy by artificial intelligence like RankBrain, it will be easier to reach the information you want to investigate.