Profile cover photo
Profile photo
Aaron Bradley
Knowledge graph strategist
Knowledge graph strategist


Post has attachment
"claimReviewed": " has made a significant contribution to building a web of linked data"

My deck for Connected Data London 2018, in which I take a look at the past, present and possible future of

#schemaorg #structureddata #linkeddata

Post has attachment
I was told there would be pie

Today in Very Important structured data news, the call-out recipe on the Carousels page has been changed from George's Cherry Pie to Party Coffee Cake. Accordingly the carousel is no longer for best pie recipes, but best desset recipes (but it's still all George, all the time).

Seriously, there's been quite a few changes to the documentation portal over the past couple of weeks - mostly small, but with some changes more notable than the promotion of coffee cake over pie. I'll try to summarize these with an eye to identifying some actionable takeaways.

Post has attachment
Save the, eh? *

As I believe some members of this Community might have a passing familiarity with, you might be interested in hearing me talk about it if you happen to find yourself in London on Nov. 7.

The entire program is fabulous (including the two sessions on other tracks taking place during my presentation that I'll have to miss):

Also, I'm in on the 6th, so if anyone has any interest in getting together prior to the conference lemme know!

* I am Canadian

Post has attachment
"The Environmental Data Initiative (EDI) has just released an experimental implementation of the and metadata to support search engine discovery and indexing"

This post provides links to the slides and a recording of a presentation that discusses this EDI initiative.

Moving forward, it seems, EDI will generate " metadata for all data landing pages on demand."

From what I've seen so far Dataset Search has been extremely well-received by publishers of datasets and are very much on board with exposing their datasets to Google, as per this EDI example.

I've memorialized this implementation on the Github page "Listing products and tools using" (

Post has attachment
Google adds support for IPTC metadata in Google Images

"As part of a collaboration between Google, photo industry consortium CEPIC, and IPTC, the global technical standards body for the news media, you can now access rights-related image metadata in Google Images."

So, as of now, Google has "added Creator and Credit metadata whenever present to images on Google Images" with Copyright Notice metadata to be adding in the coming weeks.

I echo +Michael Andrews' sentiment about this on Twitter (

"This is cool, because it shows that search results use many vocabularies, not just (as important as that is!). Well done IPTC!"

#images #metadata #iptc #cepic #rightsmanagement

Post has attachment
"It is probably clear by now that Dataset Search is only as good as the metadata that exists on the Web pages for datasets"

A new post today on the Google AI Blog provides Google's deepest dive yet into their recently-released Dataset Search.

A few things to note for the semantically-minded.

Connecting replicas of datasets

The post describes how Google attempts to determine when two datasets are replicas of one another.

Mechanisms include use of, the same canonical dataset source, two datasets pointing to the same canonical page, or two datasets with an identical Digital Object Identifier (DOI).

Note that hyperlinking "to new, recommended DOI resolver" is queued up for 3.4 (

Reconciling to the Google Knowledge Graph

I found this section, where the post talks about Google trying "to reconcile information mentioned in the metadata fields with the items in the Knowledge Graph" particularly interesting.

TL;DR (for me at least) > When Google encounters an entity it tries to reconcile it with the Knowledge Graph.

Ranking of Dataset results

TL;DR here is that, for the time being, Google is relying on "on Google Web ranking." However, as "ranking datasets is different from ranking Web pages" they're planning on using additional signals for Dataset ranking.

Open data

"The decision to rely on open standards (, W3C DCAT, JSON-LD, etc.) for markup is intentional, as Dataset Search can only be as good as the open-data ecosystem that it supports."

The post goes onto say that " Google Dataset Search aims to support a strong open data ecosystem by encouraging" adoption of open metadata standards, further development of these standards, "[the] culture of citing data the way we cite research publications" and the development of tools to leverage dataset metadata.


Just noting that +Dan Brickley gets a well-deserved shout out in the acknowledgements here. :)

#datasetsearch #datasets #schemaorg #structureddata

Post has attachment
"Methods, systems, and apparatus, including computer programs encoded on computer storage media, for triggering rich results in response to queries"

Fascinating look at the Google patent "Rich Results Relevant to User Search Queries" from - who else? :) - the venerable +Bill Slawski.

Bill, I think your concluding thoughts with regard to the role of the Knowledge Graph are especially interesting.


Instead of a rich result, Google may have decided to show knowledge panels for books, which would be a business process decision that it could be faced with when it comes to other types of entities that it might show answers for. Here is a knowledge panel for the Book, The Sun Also Rises, which contains a lot of the things that this patent tells us a rich result would contain...

Indeed. I think one of the things we're witnessing now is Google gradually morphing from rich results/rich snippets to Knowledge Panels, as the Graph grows and information that was once restricted to rich results Google is able to percolate into Knowledge Panels.

Post has attachment
Google makes several AMP-facing tweaks to their structured data requirements for article

On 20 Sept. 2018 Google's structured data guidelines for Article ( were changed to accommodate and highlight AMP pages (Accelerated Mobile Pages), and in particular AMP stories.

* List of "enhanced features" for which an article might be eligible extends from "top stories carousel and rich result features" to "top stories carousel, host carousel, Visual stories, and rich result features (i.e. "host carousel" and "Visual stories" are new). "Visual stories" is a reference to AMP stories (
* Markup examples, previously for "an Article object" and "a Video object" are now "an AMP page with Article object" and "an AMP page with Video object".
* Expanded guidelines, including further information on AMP logo guidelines and AMP story metadata (see further details about AMP story metadata with accompanying screenshots in the comments).
* Expanded guidelines about the use of the datePublished property (the date shouldn't change, use of hour information in the timestamp recommended, value for dateModified should be more recent than that of datePublished).
* Expanded guidelines about the use of the headline property ("For AMP stories, the headline should match the text in the first cover page in the AMP Story").
* Expanded guidelines about the use of the image property ("Due to format differences in search results, the followingAdditional image guidelines only apply to general AMP pages, not AMP stories. AMP stories have different requirements for images."). Again, see comments about AMP story metadata.
* More emphasis on the fact that video objects are only supported for AMP pages (from "Video objects are not supported on non-AMP pages" to "Video objects are only supported on AMP pages", with the declaration now in bold).

#amp #schemaorg #articles #ampstories

Post has attachment
Keyword research, meet semantics

What is a "keyword ontology" you ask?

"A keyword ontology is a knowledge graph describing relationships between the keywords your target audiences frequently use in search queries and the content about the products and services you sell."

A great deck from Mathewson, Priestly and Segal of IBM - required reading, I'd even say, for search marketers who have the ambition to become semantic search marketers. :)

#keywords #ontologies #taxonomies #machinelearning

Post has attachment
Dataset Search makes its debut at Google

As per the call-out link, Google today announced that they have "launched Dataset Search, so that scientists, data journalists, data geeks, or anyone else can find the data required for their work and their stories, or simply to satisfy their intellectual curiosity."

Those who have been following the development of dataset vocabulary and Google's implementation guidelines have long wondered what sort of rich results might be generated for dataset searches in the SERPs, and now the answer has been provided: none (at least for the time being). :)

Instead Google has opted to mint a new search engine just for dataset searches:

Unsurprisingly, Data Search is fueled by structured data, based on

"To create Dataset search, we developed guidelines for dataset providers to describe their data in a way that Google (and other search engines) can better understand the content of their pages. These guidelines include salient information about datasets: who created the dataset, when it was published, how the data was collected, what the terms are for using the data, etc. We then collect and link this information, analyze where different versions of the same dataset might be, and find publications that may be describing or discussing the dataset. Our approach is based on an open standard for describing this information ( and anybody who publishes data can describe their dataset this way."

The Google specifications for Dataset markup can be found here:

This is the culmination of two years of publicly-announced development efforts in support of datasets.

23 Sept. 2016
- First dataset markup specifications published

23 Feb. 2017
- "Science datasets" renamed to simply "Datasets"

11 Jul. 2018
- Google adds beta support for tabular datasets
I don't yet know whether or not these are supported in Dataset Search

31 Jul. 2018
- Google announces support for information from datasets in its search results

Further details on each of the items above can be found in "Key Structured Data Events" (today's announcement marks the 121st entry in this table):
Wait while more posts are being loaded