Infomania

As participatory government brings an onslaught of public comments online, agencies will need the right tools to make sense of it all.

In 2008, the General Services Administration had 214 million interactions with U.S. citizens through its nine Web sites, RSS feeds, e-mail alerts and call centers.

GSA expects that number to explode in 2009, as the agency enters a new era of Web-based citizen engagement under the new administration.

"When we saw the Obama administration's Jan. 21 transparency and open government memo, we thought, 'Wow! A new day has come,' " says Teresa Nasif, director of GSA's Federal Citizen Information Center. "The whole concept of government being more participatory and asking the public for their feedback is definitely a new chapter."

GSA is not alone in its anticipation of what it means to solicit public opinion on policy and regulation. Across the federal market, agency officials are scrambling to figure out how they are going to meet the administration's requirements for participatory government.

By May 21, federal officials will have more details. The Office of Management and Budget director is expected to issue an open government directive that outlines specific actions agencies should take to create a more transparent, participatory and collaborative government.

The directive is a follow-on to President Obama's January memo, which urged agencies to solicit public feedback and to use "innovative tools, methods and systems" to collaborate with one another and constituents.

What worries federal officials most about the OMB directive is how they will manage millions of public comments in a meaningful way without tying up too many resources. They want to benefit from real-time constituent feedback, but they need to set realistic expectations so the exercise doesn't come off like a publicity stunt. Automation is the answer. Observers say business intelligence software - including data mining, decision support, reporting and Web analytics tools - will help agencies extract useful information from public comments.

"More and more of public opinion and commentary are going to be sustained through the new media: blogs, social networks, all the Web 2.0 media," says Jim Kobielus, senior analyst with Forrester Research in Alexandria, Va. "Obviously, it's not a statistically valid representative sampling of how the public feels at any point in time, but it is important intelligence to use in conjunction with formal opinion polling."

Web-based commentary can provide agencies with information on all aspects of government policy. "The issue is how to mine it, how to tap into it, and how to look at patterns and trends," Kobielus says. "The geeky parlance is to look for entities and relationships and sentiments that are expressed in that unstructured content."

GSA has launched a research project to identify software that federal agencies can use to manage citizen engagement over the Web. "We're trying to figure out what tools - such as identity management, idea ranking, blogging and commenting tools and markup tools - could be used governmentwide," Nasif says.

It doesn't make sense for every agency to come up with its own toolbox for participatory government, according to Nasif. "We have to collaborate," she says. "We'll save so much time and energy and effort if we can figure it out together."

The White House, for example, is using Google Moderator, a free software application, to allow visitors to its Web site to vote on topics. "The White House Office of New Media is out ahead of us," Nasif says. "They are setting a standard for where they are expecting us to go." Adopting software to manage and analyze citizen comments is new ground for federal agencies. Up to this point, they have been managing public discussion on their Web sites manually.

Moderators screen comments for profanity or threatening language, and bloggers answer questions online. The process doesn't get any more high-tech than simple word searches in terms of analyzing the comments they receive.

The workload can be intense. When the Transportation Security Administration began its Evolution of Security blog in January 2008, it was inundated with 700-plus comments in the first 24 hours. The blog is still popular with travelers.

For example, an April 3 post about an employee's questionable behavior while screening a suspicious box at the Lambert-St. Louis International Airport triggered 156 comments in three days.

TSA's five bloggers and two dedicated moderators screen all comments and answer questions. Lynn Dean, manager of strategy and Web communications at the agency, says being responsive to the public is a challenge. But she believes the blog is creating a better image for TSA because the agency is able to explain why its officers do what they do. "We allow people to vent," she says. "We do print complaints. We look at them, and we share them. Everyone is learning from it. Before, people had no way to tell us what they were thinking."

Comments and suggestions on the blog have helped TSA improve airport security, according to Dean. And she encourages other agencies to be fearless.

"In the end, blogging will improve what you do because you're getting real, authentic feedback," she says.

Structured vs. Unstructured The challenge for federal agencies is that free-flowing comments posted to the Web are harder to analyze than structured data, such as yes or no answers on online forms.

Software exists for analyzing unstructured data, but agencies haven't used it to cull Web-based comments. For example, SAS Federal produces textmining software that allows users to comb through unstructured text, such as e-mails and blog content, to look for patterns. The program has been used to analyze prisoners' e-mails, physicians' notes and warranty claims.

"What agencies are saying is that they want to understand the themes in the comments," says John Stultz, systems engineering manager at SAS Federal. "They want to do content categorization."

The problem is people use different words to describe the same topic, and they misspell words. Without software that can analyze unstructured data, Stultz predicts agencies will be swamped. "Unless you have some way to do content categorization, the value of that information is going to go out the door. Nobody is going to have the ability to read all that information. Why collect it if you can't analyze it?" he says. Robert Dolan, global solution executive with IBM Cognos' public sector, says many agencies already use business intelligence software to measure operational and financial performance and to meet federal reporting requirements.

"Agencies are looking at budgets and how budgets tie into program results," Dolan says. "Things that are easily measured - that's where we see the greatest use of [business intelligence] technology right now."

Today's business intelligence software can't handle unstructured data well. Many vendors, including IBM, are working in their labs to improve text classification of unstructured data and to determine basics such as whether a comment is positive or negative.

One option is to solicit comments in a structured way, such as through an online form, survey or poll. This approach allows agencies to gather pre-categorized information that can be fed into business intelligence software, which can quickly churn out reports. Still, agencies will want to solicit some open comments with their online surveys because it will help them interpret responses, observers say.

"Keep the unstructured content around because it does add richness for the humans who want to delve into it," advises Don Campbell, chief technology officer for business intelligence and performance management at IBM. "But in order for the system to be more accurate in terms of rolling up and aggregating what is trying to be said, it needs some sort of form that allows a structure to be built into it."

Decision support software can help agencies bridge the gap between structured and unstructured data. The software walks users through a decision-making process in which they have to choose one option over another, similar to the way an eye doctor asks patients which of two lenses is better.

"The classic example of how it's being used in government is prioritizing budgets," says Rich Dougherty, CEO at decision-support software manufacturer Expert Choice. Agencies can use the software to walk citizens through policyrelated trade-offs and help them prioritize. For example, Dougherty says, an agency might ask citizens to rank their interest in various forms of power generation - solar, wind, nuclear and coal - but also would ask them to indicate how much additional money they would be willing to pay for electricity from each source.

"The philosophy is to nudge respondents down a path so you get them to give you very focused input and . . . you give them feedback about themselves," he explains. "The idea for these Web-based forums for citizens is to give input to the government, but the government first provides categories of choices rather than a flat text box."

Voice of the Customer

Federal agencies are new at gathering and analyzing public comments, but industry is not. Many traditional and Web-based businesses have found ways to manage feedback by combining automated and manual processes, structured and unstructured data.

Angie's List, for example, an Indianapolis- based Web site offering firstperson reviews of contractors, service providers and doctors, has 750,000 users nationwide. The site now receives 40,000 consumer-written reviews per month, double the rate of the previous year, says Gary Rush, the company's vice president of technology. Handling this volume of reviews creates several data management challenges for the 50-person IT shop at Angie's List. As the database of reviews has grown, so has the amount of time it takes to test or deploy features and the cost of ballooning storage requirements.

"You've got to have good database administrators in order to manage that data. The cost of an error becomes a lot greater," Rush says. "If somebody puts a storage procedure on the database that's malformed . . . that can bring the database to its knees."

The site uses automated and manual processes to check reviews for inappropriate language. "We have a set of algorithms that are scanning for that kind of language, which is our first line of defense, and we do some automated repair of content, but we also have a human review," Rush says.

In terms of data mining, Angie's List monitors the most popular search categories among its members as well as the companies getting the most reviews, scanning reviews to look for patterns of fraud.

"There are a lot of ways that people like to get at this data besides raw keyword search, whether it's time of submission or geography or frequency of submission or category of submission," he says.

Rush's advice to federal agencies is to make it easy to submit information over the Web. And he urges them to have plans to address storage, scalability and reliability: "Ignore any of those at your peril."

Blue Cross Blue Shield of Florida has a similar challenge with the volume of customer information it handles.

The company collects both structured and unstructured data - information ranging from rankings of health care priorities to reviews of new programs - from its 4.2 million members via the Web.

The health insurance provider uses decision support software to poll members online, and its social network - dubbed The Innovation Community - to gather customer feedback.

Carey Hepler, innovation director at Blue Cross Blue Shield of Florida, says consumers have strong opinions about health care services, just as they do about government policies. The company conducted an online survey of 900 customers in November 2007 to rank their health care priorities. "It's very easy for customers to say this is more important than that and how much more important it is," he says.

"Once we have the data, we can segment it by age segments, racial segments. . . . We can prioritize the needs of various segments, and then we use that information to create solutions."

With its online surveys, Blue Cross Blue Shield of Florida gathers only structured data that is fed into a database for analysis and reporting. To generate comments on its decisions or ideas, the company turns to its social network, which averages 1,000 comments per month. The feedback "can be very humbling at times, but it's very powerful," Hepler says. "It seems to defuse a lot of the internal conflict about what customers think of an idea."

Social Media Challenges

Social media specialists say agencies face several challenges in adopting customer feedback techniques like those used by Blue Cross Blue Shield of Florida and Angie's List.

One is legal restrictions. When GSA, for example, sought to team up with social networking sites to boost interaction among agencies and the public, it took nine months to negotiate standard terms of services with providers because the government is bound by myriad federal regulatory requirements. The agreements with YouTube, Flickr, Vimeo and blip.tv finally were signed in March.

"Numerous agencies were involved," Nasif says. "Security people, attorneys were involved. We worked with the White House Office of New Media and the Federal CIO Council Best Practices Subcommittee on Web 2.0." But she says the lengthy negotiations were worthwhile because of the opportunity for government to reach so many constituents through these popular Web channels.

"We were glad that social media providers thought that the government was an important component of their content," Nasif says. "We were excited because some of them have such huge audiences. If you're in communications in the government, why wouldn't you want to walk into a room with 175 million people?"

Nasif expects each agency will appoint a director of new media who will be responsible for the administration's push for a transparent and open government.

"Still, there is a lot of hesitation. People are asking: Is this really serious? Is this a valid way of reaching the public?" Nasif says. "All you can do is go out there and do it. The memo helps. The May 21 action plan . . . will certainly accelerate all of this."

But it's unlikely that agencies will get additional resources to adopt the participatory government directive, according to Nasif. "Program managers are going to have to evaluate how they will be adding this to their arsenal of tools to meet their mission mandate, and then they'll have to decide to let something else go," she says.

Getting rid of old ways of doing business is key. Otherwise, interacting via social media ends up being another chore for government workers. This is an area where GSA has not done a good job, Nasif admits. "We're still doing all the traditional media, and we're doing the new media, which is one of the reasons we feel harried," she says.

Duplication of effort is a problem for intelligence agencies as well with their popular Intelink suite of social media applications, which includes wiki, blogging, social bookmarking, video sharing, photo sharing, chat and other software.

With 100,000 users across unclassified, Secret and Top Secret networks, Intelink is a quagmire of information tools.

"We're suffering user fatigue . . . from bringing in far too many tools," says Chris Rasmussen, the Intellipedia wiki guru whose self-chosen title is "purple intelligence and mash-up evangelist" at the National Geospatial- Intelligence Agency. "People are having to [enter information] in two or three places. . . . We're adding and adding tools, and we're not taking anything away."

Rasmussen says federal officials should come up with a strategic plan before they start installing social media tools. "A lot of these tools are brought in because they're faddish. Let's install a wiki, or let's do a blog," he says. "That's cool, but you need to plan out what [you] are trying to do. What are [you] trying to replace. What are [you] going to reward to get people to use these new tools and new transparent processes?"

But many agree that the biggest challenge in adopting social media is cultural, not technological. "The tech solutions are the easy part," GSA's Nasif says, adding it's getting people to change the way they do business and to collaborate that will be difficult. Eventually, she says, federal employees probably will enjoy engaging with citizens through the Web. But as federal program managers venture into the online world of interactive government, Nasif offers this piece of wisdom: "If it isn't fun, it isn't social media."

Carolyn Duffy Marsan is a high-tech business reporter in Indianapolis who has covered the federal IT market since 1987.