Google’s RankBrain and Entity Analysis

Kristine Schachinger at Search Engine Land tackles the relationship between Google’s RankBrain use machine learning and the entity analysis Google developed in the Knowledge Graph.

How RankBrain Changes Entity Search (Oct 29)

Key section below  – but you will really want to read the entire article yourself.

So while Google can understand known entities and relationships via data definitions, distance and machine learning, it cannot yet understand natural (human) language. It also cannot easily interpret attribute association without additional clarification when those relationships in Google’s repository are weakly correlated or nonexistent. This clarification is often a result of additional user input.

Of course, Google can learn many of these definitions and relationships over time if enough people search for a set of terms. This is where machine learning (RankBrain) comes into the mix. Instead of the user refining query sets, the machine makes a best guess based on the user’s perceived intent.

Google ranks results with RankBrain AI

AI has arrived at Google, after years of  corporate acquisitions and experimental work with machine learning. It’s called RankBrain and it does the following:

RankBrain uses artificial intelligence to embed vast amounts of written language into mathematical entities — called vectors — that the computer can understand. If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries.

Already RankBrain  is the third most important ranking factor.

See Google Turning Its Lucrative Web Search Over to AI Machines by Jack Clark, Bloomberg, (Oct 26) for a video and description.

Danny Sullivan provides background and details in  FAQ: All About The New Google RankBrain Algorithm [Search Engine Land, Oct 28] t RankBrian is not a new search algorithm: it is one more component (albeit important)  the overall Hummingbird search algorithm introduced a couple of years ago.

Sullivan refers to the Bloomberg article and hazards a guess that the other two top ranking signals being used by Google are 1)  links – still, in spite of problems with these, and 2) words – ie matching on the search terms. Sullivan also mentions that Google has been expanding words for several years – word variants and related words – and that these fit into selecting and ranking results.  Google also employed entity analysis in providing answers through the Knowledge Graph.

Of interest: 15% of the 300 billion queries Google handles each day are new, and being new may lead to some adjustments to algorithms by the staff of search analysts .

Among those can be complex, multi-word queries, also called “long-tail” queries. RankBrain is designed to help better interpret those queries and effectively translate them, behind the scenes in a way, to find the best pages for the searcher.

For those wishing to know more about how RankBrain works with “word vectors”, Sullivan points to a couple of papers.

Greg Finn at Search Engine Land provides another synopsis – AI has officially made it’s way into Google’s search algorithm, here’s what you should know.

Can Bing be far behind in also employing AI for its search results?

Yahoo Google deal

Yahoo, if allowed, will partner again with Google for some search results and ads. Arrangement, which started on October 1, is for three years. Seems fraught with potential difficulties not the least of which getting approval from the US Department of Justice.

Yahoo & Google Together Again In New Search Deal, Danny Sullivan, Search Engine Land (Oct 20)

Would apply to the following countries: United States, Canada, Hong Kong, Taiwan, Singapore, Thailand, Vietnam, Philippines, Indonesia, Malaysia, India, Middle East, Africa, Mexico, Argentina, Brazil, Colombia, Chile, Venezuela, Peru, Australia and New Zealand.

Singhal on Google

Google’s search chief Amit Singhal described a Google search future that is mainly mobile. He spoke about – Now on Tap — a facility “which is being released as part of the latest Android mobile operating system – lets users get related information about whatever is on their handset’s screen with a single button press.”

Google search chief Amit Singhal looks to the future, Leo Kelion, BBC (Sept 29)

Image recognition is another area receiving Google’s attention. Good

Google Health Search

There is no shame today in “googling” a health condition – results are often from the top medical sites – and this is about to get better.

Google To Double The Number Of Health Conditions In Search Results Barry Schwartz, Search Engine Land (Sept 3).

Google announced they are more than doubling their health conditions database, so that when you search for health or medical topics in Google, you are more likely to find factual medical data on that condition.

I checked congestive heart failure on and was impressed at the quality of results from US and Canadian sites in the three pages. But – searchers still need to be careful to distinguish the authoritative sites from the product sites, and to thoroughly search the authoritative ones..