Eric Enge explains Google’s RankBrain and machine learning in Why Google Uses RankBrain – Here’s Why #65 (Stone Temple, Apr 25). Two examples given illustrate how Google now understands “why is” and “without” both of which would have been ignored in the past. There are many more examples on improved search results in the Stone Temple report.
Smartphones have changed the search experience. Adam Dorfman in Search Engine Land shows that As search changes, Google changes.
“But, seemingly overnight, everything changed. Now, searching means utilizing a wide range of interfaces, including GPS devices, wearables, smart objects such as Amazon Echo and operating systems such as iOS and Android. Oh, and we’re not just lounging on our sofas at home when we search. We’re searching on the go.”
While mobile is where the action is today, the author identifies some areas where Google.com still excels.
Google has quality raters – actual people who look at search results and rate quality. Their guidebook has been updated to give more emphasis, says Jennifer Slegg, to emphsize local, and expertise, authoritativeness, trustworthiness. Slegg provides a detailed account of the 146 pages in Updated Google Quality Rater Guidelines: No More Supplementary Content, Emphasis on E-A-T, The SEM Post (April 4)
The keen searcher should know the factors on a web page that affect ranking – specifically the metadata. This article explains how title and meta description.
Meta Tags: Writing Meta Data for Search Results & CTRs, Aimee Beck, SEM Post (Mar 29)
Special message to the SEM people:
I’ve said (countless times) that these two SEO elements can make or break the visibility of your web pages. That’s because the search engine results pages are your first opportunity for conversion – to get the click. And now, as the search landscape changes yet again, getting that click not only means getting traffic to your site, it also means a ranking boost … which means even more traffic to your site.
Google’s Search Quality Senior Strategist, Andrey Lipattsev, has confirmed that the top factors for ranking are content, links to your site, and RankBrain (the machine learning algorithms). There’s an hour-long video to go with this stunning revelation.
Now we know: Here are Google’s top 3 search ranking factors, Barrie Schwartz, Search Engine Land (Mar 24)
Do algorithms that are supposed to suss out meaning in content and in queries have biases? We ccould agree with Matthew Reidsma that anything created by a human will have a bias – to some degree. He examined Topic Explorer developed at ProQuest to direct searchers to the relevant subject domain where they might obtain background, context, and more relevant articles. But, sometimes the matching of user intent and algorithm rules misfires. Is it actually bias in the algorithm to pick one thing over another? Is it a weakness in the algorithm in deriving meaning? Perhaps it is bias in the content.
Algorithmic bias in library discovery systems does make the point that we should not have blind faith in the quality of search results from any search engine, not even library discovery systems. The words we use make a difference, as does the content being searched.
Long ago everyone wanted to know how their web page ranked in Google’s scoring system, and Google obliged by providing ways to obtain the PageRank. Google still has elaborate algorithms that involve hundreds of factors in assessing authenticity or goodness of a page or site, but it will stop providing the numeric pagerank score through a toolbar or any other means.
RIP Google PageRank score: A retrospective on how it ruined the web, Danny Sullivan, Search Engine Land (Mar 9)
The absence of PageRank scores may cause some to seek alternatives, estimates from third parties about how authoritative pages might be. These remain, of course, just guesses. Only Google itself knows the actual PageRank score for a page — and as can’t be said enough, the score alone isn’t the only thing that determines if a page ranks well.
RankBrain is Google’s new ranking algorithm which uses machine learning over time to “understand” searchers’ queries and retrieve better matches. Stone Temple Consulting conducted a study to see if results have actually improved.
RankBrain: A Study to Measure Its Impact, Eric Enge, Stone Temple Consulting (Mar 9)
This cannot be proved definitively, but indications are that Google seems to be handling ambiguities better (eg is coach a bus, a purse, a person?) and can recognize entities such as “The Office”. Google has also inproved in handling those natural language questions of what is, who is, where is, convert – and others.
Article has several illustrative examples.
Semantic web suggests that a web search engine can “understand” content enough to match that to an understanding of searcher needs. Today, the concepts or technologies of the semantic web are being applied in the development of databases and huge knowledge vaults..
What’s New With the Semantic Web by Donald Hawkins (Information Today, March) notes points from a keynote address by Matt Turner (CTO for media and entertainment at MarkLogic).
The startling paragraph indicates that semantic linkages will change publishing and information delivery:
“We have moved from publishing form-based prod ucts to a dedicated infrastructure in which products are created from databases of information. The semantic web plays a major role in such activities. Semantics are a good way to organize disparate data, so we must think about a new class of information. Data helps us understand customers, authors, and specialized subject areas. The use of semantic data is one of the biggest changes in our industry.”
Several major digitizing projects and their data models are mentioned.
Quick introduction to use of machine learning by Google and implications for SEO in this video at Stone Temple Consulting – Why Machine Learning Is Revolutionizing Search (Feb 29)
View it in combination with the posting – The Machine Learning Revolution: How it Works and its Impact on SEO – provides a very high level view of what machine learning involves in gathering data and seeing patterns.