Librarians need to understand what the semantic web is and how to use it, but this can be challenging. While the promise of the semantic web has existed for over a decade, to the uninitiated there may not seem to be many implementations that are accessible to the average person.
By Margaret Heller
Librarians need
to understand what the semantic web is and how to use it, but this can be
challenging. While the promise of the semantic web has existed for over a
decade, to the uninitiated there may not seem to be many implementations that
are accessible to the average person.
One
implementation that most people use daily is Facebook’s Open Graph Protocol,
which is their version of the semantic web. This is a useful example to
illustrate the ideas behind the semantic web and linked data. Libraries and
other cultural institutions want and need to make their data open, and
Facebook’s openness is highly questionable, so it will also illustrate some of
the potential problems with linked data that isn’t open. There is much great
work being done in the library world with the semantic web and linked data,
which will be addressed in more detail in further posts.
Image Attribute: Facebook's Open Graph Flow
The Semantic Web
and Linked Data
The “semantic
web” describes a web where data is understood by computers in some of the same
ways humans understand it. Tim Berners-Lee illustrates this wonderfully in his
2001 Scientific American article with a future in which the
diagnosis of a family member with cancer is made easier by the smart device
which can find the most appropriate specialist in a convenient location at a
convenient time, with very little work on the part of the searcher. This is
only possible, however, when data is semantically meaningful. Open hours for a
doctor (or a library) written on a website mean something to a human, but very
little to a computer. Once those hours are structured in a way that can be made
meaningful, the computer can tell you if the doctor’s office is open–and if it
has access to your calendar, what you have to cancel to go there.
Linking data
takes this implementation a step further and makes it possible to connect data,
to avoid,as
the W3C says “a sheer collection of datasets”. Berners-Lee outlines
the steps that need to be followed to make linked data in a 2006
post, namely to use uniform resource indicators (URIs) as names, to present
those URIs in the hypertext protocol, use a
standard format such as RDF to present useful information, and link to
additional URIs with related information. A 2010 follow-up points out that to
be linked open data, the data must be presented with a license
that allows free unimpeded use, such as the Creative Commons CC-BY license.
Such data doesn’t have to be structured in any particular way as long as it’s
open. He says that “…you get one (big!) star if the information has been made
public at all, even if it is a photo of a scan of a fax of a table — if it has
an open licence.” But “five-star” linked open data meets all of the above
requirements as well.
Facebook’s Open
Graph Protocol
Moving into a
different world, let’s consider what the semantic web and linked data look like
at Facebook. First, it is interesting to consider what Facebook was before it
was semantic. When Facebook first started in 2005, you could make a list of
things you “liked”. You might have said you “liked” the movie Clueless and
“liked” running, but these were just lists that would let others in your
college classes know a few facts about you next time you saw them in class or
at a party. In theory you could use these lists to find others that shared your
interests, but this required a person to understand what interests matched each
other.
But starting in
2010 these “likes” took on a real semantic meaning. Suddenly “liking” the movie
Clueless meant that, among other things, the owners of the “Clueless” identity
on Facebook could directly send you marketing announcements. In addition, you
could “like” content outside of Facebook completely as long as that website
used the correct markup on the page to speak to Facebook, and thus link
together content with people. Unlike Facebook’s earlier scheme of Beacon,
it was easier to understand how you were exposing yourself to advertisers and
to control privacy and sharing, though this still left people troubled.
In late
2011/early 2012 Facebook opened up this system even more to third party
developers, which went along with the new Facebook Timeline. Now any person
could perform any verb with any application. So “Margaret read a book on
Goodreads” or “Margaret listened to a song on Spotify”–real world actions–turn
into semantically meaningful statements on my Facebook Timeline. As long as the
user authenticates the application, the application can access the necessary
information to grab the information about the object from the webpage and show
the user’s interaction with it.
Developing for
the Open Graph
The Open Graph
protocol was developed based on the idea of the “social graph”, which
represents the connections between people and the types of relationships they
have with each other. In the Facebook universe, this includes the relationships
people have with other types of entities, such as media, products, and
companies. It was developed by Facebook to make a quick and easy way for
websites to include semantically meaningful data. It is based on the standard
RDF specification for linked data and includes basic and optional metadata, as
well as different types of structured data about objects, of which music and
videos are the most well-defined.
To see the Open
Graph in action, simply replace “www” with “graph” at the beginning of any
Facebook page. For instance, let’s take a look at my own library’s information
at http://graph.facebook.com/rebeccacrownlibrary.
You can see that this page describes a library, and get our phone number, physical
location, and open hours. Most important, a computer viewing this page can
understand this information. For complete details, see
the Graph API documentation–even for non-developers this is interesting;
for instance, find out how to get the URL for your current profile picture to
embed in other sites. To get access to this information, you can use various
methods, including the Facebook
Query Language.
Of course, you
only get access to this information if it’s explicitly made public by the page.
For anything beyond that, applications must use authentication in
order to access more. Linking information from outside of Facebook is one way
only–you can’t pull very much at all out of Facebook into the open web. Note
that, for instance, Google searches will pull up only basic information from a
Facebook page rather than any content that page has posted.
Outside of
Facebook–How “Open” is the Open Graph?
It is precisely
this closed effect that has a lot of people worried about Facebook’s
implementation of the semantic web. Brad Fitzpatrick described the problems in
2007 inherent in implementations of the “social graph” on the web, which was
that standards were quirky, non-interoperable, and usually completely walled
off. The solution would be a Social Graph API that would create a social graph
outside of any one company and belonging to all. This would allow people to
find friends and connections without signing up for additional services or
relying on Facebook or any other company. Fitzpatrick did later create a
Social Graph API, which Google recently pulled out of their products. Some of the
problems of an open social graph are familiar to librarians: people are hesitant
to share too much information with just anyone about with whom they associate,
what they like, and what they think (Prodromou). The great boon for advertisers
in social networking services is that inside walled gardens with reasonable
privacy controls is that people are willing to share much more information.
Thus the walled garden of Facebook, inaccessible to Google, means that that
valuable social data is inaccessible. It is perhaps not coincidental that
around the same time Google stopped supporting the open Social Graph API that
they released the API for their own social networking service Google Plus.
Concerns with
the Open Graph remain that it is not actually open, and in particular that it
uses the open standard of RDF to ingest but not share content (Turenhout). The
Open Graph Protocol website states that a variety of big websites are
publishing websites with Open Graph markup and it is ingested by Facebook (of
course), Google, and mixi. It remains unclear how much this particular
standard will be adopted outside of Facebook.
Conclusion
Whether or not
you think you have any idea what linked data is, any time you click a
“like” button on a website or sign up for a social sharing app in Facebook, you
are participating in the semantic web. But every time that data link goes
behind a Facebook wall, it fails in being open linked data.
Just as librarians have always worked to keep the world’s knowledge available
to all, we must continue to ensure that potentially important linked data is
kept open as well–and with no commercial motive. The LODLAM Summit had outlined and continues to work on what linked open data looks like for
libraries, archives, and museums. The W3C Library Linked Data Incubator Group released
its final report in fall 2011, which provides a thorough overview of the
roles and responsibilities of libraries in the world of linked open data. There
is a lot of possibility around this area right now, and the future openness of
the world wide web may very well depend on action taken right now.
In a future
post, we will examine some specific examples of work being done in the library
world around the semantic web and linked data.
About The Author:
Margaret Heller, Digital Services Librarian at
Loyola University Chicago
Works Cited
Axon, Samuel.
“Facebook’s Open Graph Personalizes the Web.” Mashable, April 21,
2010.http://mashable.com/2010/04/21/facebook-open-graph/.
Berners-Lee,
Tim, James Hendler, and Ora Lassila. “The Semantic Web.” Scientific American
284, no. 5 (May 2001): 34. doi: 10.1038/scientificamerican0501-34
Berners-Lee,
Tim. “Linked Data.” Design Issues, July 27, 2006.http://www.w3.org/DesignIssues/LinkedData.html.
Fitzpatrick,
Brad, and David Recordon. “Thoughts on the Social Graph.” Bradfitz.com, August
17, 2007. http://bradfitz.com/social-graph-problem/.
Geron, Tomio. “Facebook Expands Open Graph To 60 New Apps, Many More
Coming.” Forbes.com(January 18, 2011): 20.
Giles, Jim. “If
Facebook Likes the Semantic Web, You’ll Love It.” New Scientist,
July 31, 2010.
Iskold, Alex.
“Social Graph: Concepts and Issues.” Read Write Web, September 11,
2007.http://www.readwriteweb.com/archives/social_graph_concepts_and_issues.php.
Mitchell, Jon.
“Google Plus Releases APIs for Search, +1s and Comments.” Read Write
Web, October
4, 2011.http://www.readwriteweb.com/archives/google_plus_releases_apis_for_search_1s_and_commen.php.
Prodromou, Evan.
“On the Social Graph API.” Evan Prodromou: His Life and Times,
February 21, 2012. http://evanprodromou.name/2012/02/21/on-the-social-graph-api/.
Turenhout,
Ryanne. “Harry Halpin on the Hidden History of the ‘Like’ Button.” Institute of
Network Cultures, March 10, 2012. http://networkcultures.org/wpmu/unlikeus/2012/03/10/harry-halpin-on-the-hidden-history-of-the-like-button/.
This work is
licensed under a Creative Commons
Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Based on a work at acrl.ala.org/techconnect.
