New DG Team Pursues eRulemaking Research:
Cornell Researchers Extend Earlier Efforts
Karen Heyman
For the DGRC

New DG Team Pursues e-Rulemaking Research:
	Researchers profile: FARINA, C. CARDIE, C. SHULMAN, S. HOVY, E. Project profile: Collaborative Research: Language Processing Technology for Electronic Rulemaking e-Rulemaking Research Group

The 7th Annual International Conference on Digital Government Research, dg.o 2006. Dr. Stuart W. Shulman, Director of Qualitative Data Analysis Program at the University of Pittsburgh, will chair a workshop on Wedensday afternoon, May 24, 2006, on "eRulemaking at the Crossroads."

Digital Government researcher and political scientist Stuart Shulman (University of Pittsburgh) is considered by DG colleagues to be the founder of the e-Rulemaking research community. While his own efforts date back to 1999, since January 2003 he has been working with sociologist Stephen Zavestoski (University of San Francisco) and computer scientists Eduard Hovy (USC-ISI) and Jamie Callan (Carnegie Mellon University) on the challenge of sorting through the sometimes hundreds of thousands of letters citizens and interest groups send in to comment on proposed federal regulations.

Recently, Cornell researchers, law professor Cynthia R. Farina, computer scientist Claire Cardie, legal informatics expert Thomas Bruce, and organizational behavior theorist Erica Wagner, received a major DG grant that will, in Farina's words, "build on the work Stuart and his group have been doing for years. You can't overstate the value of what they've done and continue to do."

One difference between the two projects is where in the commenting process they will focus. Shulman's "e-Rulemaking Research Group" looks at what happens after comments are received, due to the expressed desire of regulators for a way to ease the burden of sorting through so many comments (http://erulemaking.ucsur.pitt.edu).

Although the Cornell team is also working on comment sorting and analysis, they will additionally concentrate on an earlier step, the interface a would-be commentator first encounters on Regulations.gov. Perhaps, they hope, they can improve interface design in ways that would encourage more pertinent comments in the first place, thus making the next phase less challenging.

As a professor of administrative law, Farina feels that many citizens, even 2nd year law students, don't understand that "the regulatory process is different from the political process." In our system of representative government, a representative will listen to the will of the people, noting how many constituents write in about particular issues. Sheer numbers alone may be enough to persuade a representative which way to vote. However, if conscience dictates, he or she may take the political risk of voting against what the majority of voters appear to want.

Regulators must work out the real-life details of the laws that have been passed. For those whose mandate is implementation, rather than legislation, the calculation is never as simple as counting up how many yeses vs. how many nos. In order to be persuaded, they need to know, in much greater depth, precisely why those opposed to a proposed rule fear it might impose economic burdens, possibly cause ecological damage, or other unintended an, unforeseen results. "Unsupported opinions are generally not useful to an agency, except for figuring out general sentiment" says Cardie, "To make a decision, an agency needs data."

Since most citizens, and even some professional interest groups, think of regulators as working exactly as legislators do, they will deluge them with postcards or e-mails for or against proposed rules. Shulman and his colleagues continue to develop ways to highlight the relatively few unique, individual comments from the masses of form letters. Cardie and her colleagues will be working on the problem as well. Cardie, Hovy, and Callan are all specialists in natural language processing; the science of creating software that can better organize human language for analysis.

The impressive search engines, with which we're all familiar, such as Google, actually illustrate the problem. They are excellent at pulling out keywords, but they cannot distinguish their context. It's still a challenge to get a computer to understand that "Bush" with a capital B is the surname of two US presidents, while "bush" with a lowercase B refers to wild land or a piece of shrubbery. Reading a sentence like, "We want Bush to save the rose bush next to the bank near the river bank," a human knows immediately that the writer wants the president to preserve a flowering plant, and knows as well that the first "bank" refers to a building. But a computer must be trained, with complex learning algorithms, recursively going over texts, before incorporating enough understanding of syntax and context to correctly interpret a sentence like, "Bush saved the bush," that any human child can grasp.

In the context of e-Rulemaking, both the original eRulemaking Research Group and the Cornell group are writing software to help computers identify the unique, individual comments that have more value to regulators than thousands of letters with the same Do/Don't boilerplate text. In addition, the Cornell group hopes to create a useful way to index and group received comments, so that someone hoping to add an additional comment could easily search to see what had already been written on particular sub-sections of proposed rules.

Latest DG News

• dg.o 2006 Convenes May 21-24, 2006   • dg.o 2006 Early Registration Ends April 10th! • dg.o 2006 Issues CFP - Tutorials • dg.o 2006 Issues CFP - Workshops • dg.o 2006 features Workshops on:    eRulemaking    GeoInformatics • dg.o 2006 features Tutorial on:    •Social Network Analysis • New DG Team Pursues eRulemaking • IEEE ISI2006 Convenes May 22-24, 2006 • eChallenges e-2006 Issues CFP • DG Research Helps Predict Urban Growth • Swapping Secrets of the Double Helix • UK and DO-Wire Launch e-Gov Best Practices wiki • DG Team Develops "Virtual Agora" for e-Gov • Mapping for Times of Crisis • Exploring Detection of Crisis Hotspots • Report: Mass eMail Campaigns Harmful • Scenario-Based Designs for Stat Studies • US, EU Explore Info Integration • DG Team Develops Digital Interpreter • DG Study Gives Teeth to FBI • Research Smooths Road for Small Businesses • DG Researchers Parsing in Tongues • e-Gov Journal Issus Call for Articles • See all news stories Contribute to dgOnline

They are considering taking on an even more daunting challenge as well: Creating software that could help rulemakers by indexing apparently unrelated statutes and executive orders to flag mandates relevant to the rules being written, to catch possible conflicts and contradictions. "So many things impinge on other things," says Cardie.

For all the optimism about how the work of both groups could make the commenting process easier and more effective for regulators and citizens, Shulman admits, "I've scaled back my expectations." Similarly, Farina cautions, "We are looking at modest steps in improving the way the interface educates people and facilitates citizen participation." What optimists about eRulemaking had not considered, Shulman now realizes, is that some groups consider it in their best interests to keep the process of rulemaking slow and tedious. If a group is strongly opposed to a proposed rule, and fears it might be approved despite their objections, then they may feel the best course they have left is to delay its implementation. In which case, they would actually prefer the current system, in which hundreds of thousands of comments are sorted by hand.

There's also another subtle issue of civics and administrative law, says Farina. Although legislators and regulators view comments differently, many interest groups be,ieve that because regulators are ultimately hired by elected officials, the sheer gross number of comments on a proposed rule should matter as much as it does when legislators are making a decision.

Nevertheless, while aware of those issues, all the involved researchers still hope to do their best to use digital technology to make the working lives of regulators easier and the outcome of the regulatory process better.