Last week, we shared about how we will soon identify individual passages of a web page to better understand how relevant a page is to a search. This will be a global change improving 7% of queries: blog.google/products/searc…
In this thread, more about how it works….
Typically, we evaluate all content on a web page to determine if it is relevant to a query. But sometimes web pages can be very long, or on multiple topics, which might dilute how parts of a page are relevant for particular queries….
With our new technology, we’ll be able to better identify and understand key passages on a web page. This will help us surface content that might otherwise not be seen as relevant when considering a page only as a whole....
This change doesn’t mean we’re indexing individual passages independently of pages. We’re still indexing pages and considering info about entire pages for ranking. But now we can also consider passages from pages as an additional ranking factor….
There’s nothing special creators need to do here. Continue to focus on great content, with all the advice we offer: google.com/webmasters/lea…
It just means in some cases, we may now do a better job of surfacing content, no work required on the part of creators.
Update: passage ranking launched yesterday afternoon Pacific Time for queries in the US in English. It will come for more countries in English in the near future, then to other countries and languages after that. We'll update this thread as those further launches happen.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Have you ever searched for something in Google Images and sometimes found the results dominated by the same image? To help, we made an improvement to Image Search to reduce duplicate images, so that we can display others that are relevant yet visually distinct....
There are also times when a query can relate to different things. For example, [jaguar] could mean either the animal or the car. Sometimes, one meaning can dominate image results. Our improvement works to show images for alternative meanings, when that seems helpful....
As a reminder, there are also buttons that may appear atop image search results that allow anyone to refine a query to a particular meaning, such as to easily focus images to be about jaguars, the animals....
We are currently working to resolve two separate indexing issues that have impacted some URLs. One is with mobile-indexing. The other is with canonicalization, how we detect and handle duplicate content. In either case, pages might not be indexed....
If a previously indexed page has gone, it might be the mobile-indexing issue, where we’re failing to select any page at all to index. If the canonical issue is involved, URL Inspector may show the URL as a duplicate & the Google-selected canonical will be different from it….
There’s no action to take with these issues on the part of site owners. We apologize for the issues here and are working rapidly to resolve them. We’ll update this thread as each is corrected.
It was noticed today hateful memes appear in image results for “jewish baby strollers.” We apologize. These don’t reflect our opinions. We try to show content matching all key terms searched for, as people normally want. But for “data voids” like this, it can be problematic...
For “baby strollers”, there’s lots of helpful content. For this, there’s not. That’s not surprising. It’s not likely a topic normally searched for, nor an actual product that’s marketed. There’s a “void” of good content to surface that matches what was asked for…
That’s the explanation of why these appear. It’s not meant as an excuse. We’ve done considerable work with improving data void situations & finding systematic improvements. We’ll look at this situation to see how we can further improve. Our apologies again for the concerns here.
When searching on Google, by default we automatically also search for any synonyms & related words to what was entered, as this is often helpful. Yesterday, our systems that do this raised concerns we were intentionally making some ideological statement. This was not the case….
If Google only returned documents that contained the exact words entered in a search, we would miss displaying relevant content. For example, “change” as used by a searcher might fail to match documents that talk about “replace” or “exchange” or “adjust” as written by a creator…
This type of query expansion happens regularly. But in some cases, we detect the top results returned may be further away in relationship to the original terms entered. In such cases, we automatically label that a broader expansion is happening & provide an override option….
We’re aware an image for Sir Winston Churchill is missing from his Knowledge Graph entry on Google. We apologise for any concern. This was not purposeful & will be resolved. Images in such panels are automatically created & updated. During an update, they can briefly disappear….
If a Knowledge Graph image is missing due to an update, the subject will be named but lack an image in anything automatically generated from the Knowledge Graph. This is why Churchill’s image (but not his name) is missing from some lists. It is temporary & will be resolved….
The image issue only applies to Knowledge Graph-generated information. Many images of Churchill are easily found through both web and image searching on Google….
Google Search has never seen as many searches for a single topic continue over a sustained period as is happening now with COVID-19. Many searches are for news about what’s happening in local areas, such as sheltering updates or the latest on testing. Here’s how we’re helping….
What is local news content? It can be material produced by publishers like local newspapers, that we understand specifically serve particular cities or regions. It can also be content we identify as relevant to a particular area, even if written by those outside the area....
Last month, we made a change to help local content surface better within our Top Stories box. In addition, we added a new “Local news” box to show local news content in response to coronavirus-related searches: