Making Semantic Web connections

Image: Semantic Web technologiesLinked Data technology can change the business of enterprise data management.

Download Technology Forecast: Spring 09.

Imagine you’re a retailer a few years from now, assessing a site for a new store that will sell golf equipment and apparel. Before you decide, you want to develop scenarios about how the store might perform; you also want to examine the potential performance of several new stores when some existing stores are closed at the same time.

You have all the information you need. You know the site, its dimensions, and the planned inventory for the new store. Public data—including demographics, regional economic statistics and forecasts, and locations of competitors—are available. The information exists in different formats at various Internet sites, but that’s not a problem because your company has adopted the Linked Data federation method. This set of techniques allows data to remain in native form, but to be exposed and blended at Web scale. This method taps into a larger number of internal and external sources than otherwise would be possible and recasts them in a common format.

Based on emerging Semantic Web standards, Linked Data technologies allow you to refine your golf store scenarios by calibrating for age distribution, per capita income, and other factors by census tract or even block group—all with data extracted from disparate sources. (See Figure 1)


Graph: A sample of some retail store information in graph form

Figure 1: A sample of some retail store information in graph form
Linked Data uses a flexible graph format to connect one data element to another from disparate sources. In this example, external data are connected from DBpedia to the rest.
Source: PwC, 2009

The disparate data feed into a mashup—a Web application with highly configurable data display capabilities—that updates each time you add a new store site or remove an old one. Other data in the mashup are refreshed whenever the original sources are updated. By combining various data, regardless of their format or source, you have a wide range of possibilities for greater insight and context. For example, you can use the same techniques to create information mashups as needed, not just for long-term uses such as the golf store example. Perhaps a business analyst wants to test changes in regional product purchases against local home sales and employment data to see whether a decrease in sales is due to local economic issues or is a possible harbinger of a broader shift in tastes. You would never create a formal application for this exploration, but with the Linked Data approach, you don’t need to.