Recently, a new resource appeared on the Web to help developers navigate the waters around various approaches to adding semantic markup to websites and applications. We caught up with the creators of the newly launched structured-data.org, to learn more about this project. They are:
 |
Stéphane Corlosquet, |
 |
Gregg Kellogg, |
 |
Manu Sporny, |
Q: What is “Structured Data on the Web” (the site and the concept)?
GK: We wanted to provide a place for people to learn about the different ways in which publishers can add semantic information to their web sites and applications. There is confusion in the marketplace, partly due to the introduction of schema.org, which has raised the awareness of structured data with web developers. structured-data.org is a one-stop-shop to learn about the different mechanisms available to developers to take advantage of this.
MS: As Gregg mentioned, Web Developers don’t have one Website that they can go to and review their options when it comes to Structured Data on the Web. This was leading to developer frustration – “Which one should I pick!?” has been a common question since the schema.org launch. We wanted to have a one-stop-shop for expressing data on the Web where Web developers can learn, play around with each technology, and then implement and debug using the tools listed on the website. We are trying to build bridges between the Microformats, RDFa and Microdata communities and this is a first step on that journey.
Q: Who do you see as the core audience you are trying to reach?
SC: The main audience we had in mind when starting this project was the regular Web developer who needs a tool to check his or her markup. Facebook provides its own linter which only recognizes the Open Graph protocol. Google offers its Rich Snippets testing tool, which supports microdata and some subset of RDFa using the data-vocabulary.org schema as well as some microformats. Bing is in the process of building their own testing tool, which most likely will also only support a specific set of syntaxes and schemas. We felt the need for a unified, non-biased testing tool to put in the hands of web developers willing to build webpages that are supported by the various search engines, Facebook, etc.
MS: Web authors and Web developers – they’re the ones that are going to make these technologies succeed or fail. We are trying to get the tutorials and tools that they need into their hands as quickly as possible.
Q: What was the impetus for the creation of structured-data.org and once you were down the path, did the vision of what it can be/should be change?
SC: It started with the structured data linter. I needed a way to test the pages that I build at work and during my spare time to see if the structured data was parsed properly, and what these pages could look like in search results pages. I didn’t want to have to try all the testing tools out there, and I wished there was one central service which supported all the structured data formats. I knew Gregg was working on several parsers in Ruby and he was keeping them up to date with the evolving specs, so I turned to him and asked if he’d be interested in working on such a linting service. He took on the idea and built it! While discussing where this service should live, Manu suggested to make it part of a larger initiative about Structured Data on the Web, with some educational content and links to the various formats and tools. structured-data.org was born.
MS: We spend a great deal of time listening to what Web developers are saying. We take input from anywhere that we can find it – Twitter, Google+, mailing lists, IRC – you name it, we’re listening. Over the past several months, there has been a rise in the number of folks that were wondering why they should pick RDFa over Microdata or pick Microdata over Microformats. As with any new technology cycle, this sort of confusion is normal, but it has been exacerbated by two compounding factors.
MS: The first is that the W3C HTML Working Group intends to publish both RDFa and Microdata as official specifications on top of HTML5. These two technologies cover close to 80% of the same use cases, but in different, incompatible ways. There is a rich history of this type of approach to standards-making – XSL vs. CSS, XHTML1 vs. HTML5, IPX/SPX vs. TCP/IP, the list goes on.
The other confusing factor is that Google Rich Snippets first launched with Microformats and RDFa support – adding Microdata later on in the process. It then announces schema.org, which only supports Microdata, but with the statement that they /may/ support RDFa in the future. Meanwhile, Facebook is supporting RDFa with their launch of the Open Graph Protocol. So, if you want to show up on both Facebook “Likes” and Google search result pages, you end up having to use both in your markup – which is a huge burden on Web developers.
So, we got together and said that the people that should be making this RDFa/Microdata/Microformats decision are really the Web developers. We want to make sure they make an educated decision, so we wanted to give them the tools to do so. We invited key people from the RDFa, Microdata and Microformats communities to participate – all of them have commit rights to the structured-data.org website. It’s meant to be the Geneva of structured data on the Web – a neutral website where people that are interested in Structured Data can go to learn and play without having to worry about corporate or community bias.
As for whether the vision has changed – no. It was always meant to be a neutral, non-biased resource for Web developers and authors.
Q: What resources are available now to that audience and do you have plans to add offerings?
MS: There are currently links to introductory articles and tutorials on the website. There are a few more interactive learning tools on the site like the Microdata Live website, the RDFa Live Loop service and the JSON-LD playground. There are also a set of data extractors that one can use for Microdata and RDFa. There is also the Structured Data Linter, which understands both RDFa and Microdata in the same document. We also plan to add support for Microformats as time allows. We wanted to have Microformat support done before launch, but it’s better to get most of the tools out there to Web Developers than it is to have a complete solution that is delayed by a few months.
The bit that that is new and that we’re most excited about though, is the Structured Data Linter. Gregg is really the driving force and hero behind the new tool, which is a really great addition to structured data debugging tools on the Web. He has proven that it is possible for a single person to build a one-stop RDFa/Microdata debugging tool in a few weekends.
GK: The linter.structured-data.org service was created using existing RDF and Linked Data parsing and serialization tools; this allows us to read a wide variety of formats and put them in a standardized representation (an RDF graph). We then leverage RDFa serialization tools to match on types represented in the graph to show the information in a snippet form. This allows us to create new snippet templates relatively easily, and new ones can be added at any time. The existing set we’ve provided give one view of what structured data might look like in search results, for example. The site is undergoing constant development, so expect support for new types, and improved support for existing types in the future.
The linter was created entirely using public-domain open-source software. These tools happen to be implemented in Ruby (my language of choice), but similar tools exist in all major languages. Most developers and publishers don’t really need to know anything about RDFa, Microdata, or Microformats to get started. Simply using existing examples and testing against the linter can be of great use to developers that aren’t interested in the details. You can be even more sophisticated even without data in an RDF store. Using standards in development such as R2RML, developers can use their existing relational databases to get RDF; this allows tools, such as the RDFa Writer to be used to publish that as HTML. Other tools can publish as JSON-LD or any other Linked Data or RDF format.
