Digital Design Blog

The Semantic Web We Weave

A look at the technology that may usher in a whole new wave of consumer
—and computer—experiences.

Why does Netflix keep sending me movie recommendations that are clearly more appropriate for my ex-boyfriend? Full Metal Jacket is just not my cup of tea. Why doesn’t Netflix know I dumped Simon last month?

In the not too distant future, the semantic Web will enable Web sites to do just that. For years, we designers have labored to do one thing really, really well—give users what they want. This is no small feat, given the complex, evolving and fickle needs of savvy Web users. But now, the emerging technologies behind the Semantic Web make it possible for Web applications to leverage information (sometimes personal) more intelligently, ultimately making consumers’ digital experiences more valuable and relevant.

Translate That Into Computerese
The underlying strategy of the Semantic Web is to create data and Web sites that are “machine-readable.” Why do we care if machines can read data? After all, we’re designing Web sites for people, right? If machines could comprehend the meaning of the content of a Web site, they could manipulate data in more meaningful ways, resulting more quickly and more frequently in precisely the information that the user wants.

For example, you can use Google Maps to find the nearest Starbucks. You can even use it to try to find that fabulous French restaurant whose name you forgot (if you know the general neighborhood). But what if you’re looking for something even more ambiguous? Say you suddenly remember that it’s your nephew’s birthday tomorrow, and you want to search for nearby stores that are open late and sell gifts suitable for a teenage boy who self-identifies as a geek.

And how about that breakup? On the Internet, personal information can last forever, but the details of people’s lives change. If my social networks and favorite sites were infused with semantic capabilities, I could change my relationship status to single on Facebook and it would alert all my other linked services. Netflix, knowing that Simon was the one who gave 5 stars to Saving Private Ryan and Full Metal Jacket, would stop making strong recommendations for other war movies.

So, how do machines know what data means?

The Building Blocks
There are three technological innovations that make the Semantic Web possible, plus one key factor that makes it feasible.

IDs
Machines need a unique, consistent way to identify a thing or concept. For example, if I mention “Bill Clinton,” how does a machine know that this is the same person as President William Jefferson Clinton? If I mention “President Bush,” how does it know if I mean George W. or George H. W.? People can usually tell by context, but a machine needs a unique identifier to be able to make those connections and distinctions.

New Web Standards
Web1.0 and 2.0 were built on standards like HTML, XML, and CSS. Some new standards have been developed specifically for expressing metadata and metadata relationships. Standards such as RDF, OWL, SKOS and Dublin Core are used to define the structure, usage, meaning, rights and authority of metadata, in a machine-readable way.

Ontology
This is the part that scares people most. All this really means is defining the shared classifications, attributes, relationships, and logic that will allow machines to integrate distinct data sets and extrapolate new, unexpected information from stated information. Think of it as a hyper-glossary.

Commercial Tools
The first three things listed here make the Semantic Web hypothetically possible, but to make it a reality we need commercially available software and systems that will allow people and companies to easily create, manipulate, interpret and use semantic data. Without these tools, the obstacles to adoption would be prohibitive for most organizations.

Is the Semantic Web Right for You?
Do you want to leverage the vast stores of excellent content and data that your company has to offer? Want to provide rich, relevant search results? How about personalization, customization or integration with popular scheduling, shopping, social or location-based services?

These are the kinds of features savvy Web users are already starting to expect. Soon they’ll be demanding them. You can probably fake some of these features for a while without Semantic Web technologies. This usually involves manual processes that are expensive and time consuming and quickly become unmanageable as your collection of content grows. Plus, what happens when the next new widget comes along and you absolutely have to add it to your pages to keep your audience satisfied? It will probably require a new development cycle, a whole bunch of manual coding and another change in work process to get that widget meaningfully integrated into your site. If your data were semantic-ready, you could have that widget up and running in the time it takes you to read this article.

A few years ago, the common view was that you wouldn’t be able to benefit from the Semantic Web without investing huge amounts of time and money. Now the prevailing attitude has shifted towards “a little semantics goes a long way.” Semantic solutions can be adopted incrementally, with the ability to grow as needed. For example, solutions that allow businesses to overlay semantic information on legacy data sets—without having to migrate them—reduce the barriers to entry. Data mining, based on natural language processing, makes it easier to discover rich data relationships that may already be lying dormant in your company’s content.

As with any emerging technology, there are costs and risks associated. Within a very short period of time, though, the opportunity cost is going to be even higher. When some information services allow me to quickly and easily extricate Simon’s influence from my online presence and some don’t, I’m going to go with the ones that know I’d rather watch Moulin Rouge than Black Hawk Down.

by Rachel Lovinger