Jennifer Zaino
SemanticWeb.com Contributor

The fourth of July holiday is upon us. And this year, why not celebrate democracy by checking out the open source project GovTrack.us, a site where you can follow the status of federal legislation and members of Congress?

The government, of course, puts lots of data about the activities of the U.S. Congress on the web. THOMAS, for example, provides legislative information from the Library of Congress. But much of the time, the user interfaces to sites like this are less than optimal, and the data in each exists in its own individual silo.

GovTrack.us is one graduate student's attempt to change the experience when it comes to getting information on pending legislation (and about a decade's worth of historical data on the topic), both in terms of a more accessible interface and pulling together data on the topic from multiple government sources.

Joshua Tauberer, a linguistics graduate student at the University of Pennsylvania, doesn't call GovTrack.us a semantic web application. But the site is underpinned by an RDF database. Because most of the government data out there is not available in a particular structured format, Tauberer screenscrapes a handful of government sites and converts the information to RDF format. Recently, he's also begun enabling site visitors to run SPARQL queries against the data.

Tauberer, who also maintains a large RDF data set of U.S. Census data (a billion triples) is just trying to do his part to put as much data onto the semantic web as possible.

"There's so little there now and you can't have interesting semantic web applications without data to play with. And you need data connected at its edges to other sets of data," he says. "Why open government data is useful and important is that you can combine and transform the data in revealing new ways."

For example, he's experimented with mashing up census data in RDF format with data the Federal Elections Commission publishes about campaign donors and their contributions (which he also converted to RDF) to plot the total per capita campaign contributions to one New Jersey representative from each zip code he represented and then plotted them on a map.

"So the point of that was that there are a lot of connections between government data and putting it into a semantic web format and then meshing it around is interesting," Tauberer says, "and it can be revealing and then seeing it on a map is a little bit informative, too."

Adoption Challenges

He'd like to see interesting semantic web applications come out of the RDF data sets he's put together. But he acknowledges that most of GovTrack.us' users are more interested in subscribing to subject areas and getting RSS feeds on the topics that interest them than they are in taking advantage of running SPARQL queries. Tauberer figures there are two factors at play here.

"In the existing semantic web community, it's fairly large but fairly widely distributed across the world, so not that many people actually are interested in any given domain. There are maybe two people in the world interested in the semantic world and U.S. politics and the legislative branch, so it's kind of a narrow area to begin with," he says.

But the other reason has bigger implications for the take-up of the semantic web: "On the other hand, a lot of people are interested in data about the U.S. Congress but not semantic web data and I'm trying to encourage these people to go the semantic web way. But you can't encourage it too much because there are not many good tools to deal with it and usually there is a simpler way to do what they want to do without using semantic web stuff."

He notes, for example, that he had turned some data from the SEC about who owns what publicly traded companies into the RDF format and heard from someone who wanted to play with the data-but was put off by the fact that it was too complicated.

"He thinks I should have just put into a MySQL database or something plain and obvious, no URIs," he says.

Using the Notation 3 RDF syntax, Tauberer says, "creates a layer of overheads, and you have to get the parser for the right file if you want to use that format."

A lack of tools and knowledge slow down acceptance. But Tauberer is hopeful that things will get better.

"Every new technology has to have some time before it's adopted. RDF is taking quite a while but it is actually a complicated beast, so I'm not surprised that people are not so into it."