"You can't just stop trucks on the road and ask them what they're carrying," says Digital Government researcher Genevieve Giuliano, a professor in the School of Policy, Planning, and Development (SPPD) at the University of Southern California (USC) and Director of the Metrans, the joint USC and California State University Long Beach Transportation Center.
Still, as a layperson you might think the solution must be relatively painless. Surely there are records such as invoices, shipping documents and cargo manifests; the only challenge must be putting it all together into one database.
But for once, the problem isn't information overload. It's information gaps. Although there is a great deal of documentation of state-to-state commerce, there is little for intra-state and even less for intra-county, says Giuliano. Don't assume for example, that it would be easy for a California researcher to get information about what is unloaded on a dock in Long Beach or San Pedro and where it is going.
"Customs data, surprisingly enough, you can't get," Giuliano says. "All customs data is given to third party contractors that manipulate the data; you have to buy it from them, and it's prohibitively expensive." These problems have led the research team to a different approach for estimating freight flows within a region - one that relies on more readily available secondary information sources.
Trying to create an automated system that would produce reliable freight statistics makes an ideal real-world challenge for a prototype project headed by Giuliano and fellow DG researchers, José Luis Ambite, a senior research scientist at USC's Information Sciences Institute, and Peter Gordon, a professor in SPPD.
Called Argos, the project is a general framework for dynamically composing Web services, according to Ambite. The phrase "Web services" can be a somewhat confusing term to those outside of the field. Hearing it, one may think of the professional services offered by Web design firms. It is actually the title adopted by the World Wide Web Consortium, an international standards body, to refer to the suite of protocols that allow applications to interoperate over the Internet.
The problem is actually one of generating a computational workflow, explains Ambite: How do you integrate data from many sources and process it to generate the answers you need? Accurately modeling freight, which will require combining disparate information, including data derived from secondary sources, is a perfect computational test for Argos.
"In Argos we represent each source and each data processing operation as a web service," Ambite says, "We also describe the contents of each source and the inputs and outputs of each operation using terms from a domain ontology, so that the system knows precisely what sources are relevant and what operations to use when composing a workflow."
In practical terms, this means, says Giuliano, "Instead of spending months to get one run of the model together, it should be able to be done in hours. It allows us to conduct many more simulations, to do a lot more scenario analysis. And we hope that by automating this model by using Web services, we can give regional planners a tool that can be used anywhere. We hope that by the end of three years people will have a model that can actually be used as a tool in transportation planning."
Further detailing the modeling challenge, Giuliano says, "Hard to believe, we have no good way of counting trucks. The state of practice is to send humans out, sit them in a lawn chair, and have them count trucks as they go by." The research team is using for their inputs secondary data, like census and employment statistics, as well as total air and waterborne shipments in and out of the region. A regional economic model is used to generate supplies and demands for goods. After a series of computational steps, flows of trucks on the highway network are estimated. These estimates are compared with a sample of actual observations to cross-validate the model.
Often Digital Government projects are the first collaborations ever between government officials and academics, who need to create common vocabularies and appreciate different timetables. But for Mark Griffin, Senior Regional Planner at SCAG (Southern California Association of Governments), working with researchers like Giuliano and Ambite is a familiar experience.
"At SCAG, we're encouraged to work with academics — most of the planning staff are USC or UCLA alumni, and many practitioners are faculty at the two schools," says Griffin. "I think it's beneficial to us all: We get access to conceptual work and they get exposure to practical considerations."
This current project is still in early stages, but says Griffin, "We're happy with what they've developed so far. We're intrigued and look forward to more results." Additional government advisors include the Los Angeles County Metropolitan Transportation Authority, the San Bernardino Associated Governments, and the Port of Long Beach.
|This site is maintained by the Digital Government Research Center at the University of Southern California's Information Sciences Institute.|| CONTACT POLICIES|