April 2001

Info-Nation

by Adam Throne

Author George Orwell portrayed them as the loathsome instruments of intrusive government. Science fiction writer Douglas Adams mocked them by saying they could only be found in locked basements guarded by filing cabinets and leopards. They were talking about, of course, government forms.

Paper records seem to breed in nearly every government interaction. These documents can be unwieldy to manage and time consuming to process. Many departments are now adopting new technology to speed processing and empower both civil servants and citizens to search for information online. Agencies are turning paper forms into easily retrieved electronic information, and they're adding rich media such as video into the mix of digital information.

For example, the U.S. Patent and Trade Office has cut patent application approval times in half by switching to electronic submissions structured with extensible markup language (XML). The U.S. Department of Energy is digitizing training videos, indexing them with speech-to-text technology and posting them online for greater accessibility. The U.S. Embassy in Mexico City is adopting color imaging technology to speed and improve the security of visa applications.

Patent Office Offers Alternative to Forms

The United States Patent and Trade Office, Washington DC, accepts more than 2,000 applications for patents each day. In the past, these documents, which each average 100 pages, were scanned and turned into ASCII text files with the help of optical character recognition technology. After the initial examination process, the text files were marked up and tagged with metadata so that they could be managed electronically and published online.

To save time, the Patent Office outsourced the job of tagging the documents, but the entire process was still prone to errors. "There were inaccurate filing receipts and data entry questions," says Robert Porter, the director of the agency's Office of System Development and Maintenance. "We often found missing parts when it came time for the patent office to review applications. This would cause delays."

To make matters worse, the Patent Office had experienced a 75 percent increase in its workload since five years ago, while also contending with budget cuts. The patent approval cycle could take anywhere from 24 to 30 months.

"We knew we needed something else to handle the future of patent processing," says Porter. "Law firms had been doing [the paper-based processing] for years and were swamped. Plus, [if there were] changes in the workflow, it would take even more time."

William Stryjewski, primary patent examiner, summed it up: "We needed a paperless office".

The obvious answer was to conduct as many transactions as possible electronically. The agency set a goal of conducting 80 percent of filings electronically by 2003.

The Patent Office chose S4/Text, from i4i, Toronto. The product offers a tagless editor that lets users create XML files in Microsoft Word (or other word processing or email programs) without having to know the intricacies of metatagging. Using the S4/Text, anyone collaborating on patent content can file directly to the Patent Office in digital form, so applications no longer have to be manually scanned or reworked. The resulting documents are packaged and encrypted using its own ePAVE (electronic packaging and validation engine) and received by an electronic mail room that immediately sends an acknowledgment. The manual system often took three weeks to send an acknowledgment.

According to Diane Lewis, the electronic filing system project manager for the Patent Office, the agency chose S4/Text because it made the experience as "Word-like" as possible while ensuring that users could not break the required structure or delete mandatory elements of the documents. The agency also wanted features that other systems did not support, such as tables.

"The system ensures that applicants submit Microsoft Word documents with XML tags already inserted," says Lewis. "Before that, we had to cut and paste [into templates]."

Since patent filings are legal documents, many applicants were used to using Word Perfect. Lewis says S4/Text can use an authoring tool that allows Word Perfect users to work with the same templates as Microsoft Word.

In October, when the system was launched, more than 10 percent of the attorneys and agents registered with the Patent Office signed up for the digital certificates that allow electronic filing. The agency estimates that if 40 percent sign up for the program, the department will save $25 million dollars. In addition to an accelerated approval time, applicants worldwide stand to gain from the system, which can be used universally.

"We worked with international properties offices from Japan and Europe, which, with the United States, cover 90 percent of the world's patents," says Stryjewski. He says that those organizations are considering adopting the system. "They've signed a world patent cooperation treaty that, if approved, will realize a global infrastructure for patent submission. "Right now, the process for other nations differs from that in the United States."

The system took six months to implement. "It was the largest deployment of XML around," says Michel Volpe, the founder and chief technical officer of i4i. "The up side is that submissions can now come in digitally. This reduced time from submission to approval by as much as 50 percent. We are well on our way to reaching our target goal for 2003."

Now on Video: Bomb Diffusion 101

Not everybody gets to play with nuclear weapons, but those who do had better be prepared for emergencies. It's comforting to know that the U.S. Department of Energy's Sandia National Laboratories, Albuquerque, NM, operates an Accident Response Group training program on the safe handling of nuclear weapons. As part of training, the unit offers videos showing how teams with various talents and tools can render nuclear devices safe.

It doesn't take a rocket scientist to figure out that sharing this kind of training is a good idea, so Sandia Labs decided to make the videos more easily available to employees by putting the videos on its intranet. The challenge was organizing and making the videos searchable.

After careful review, Sandia selected Convera, San Jose, CA, to provide a means of archiving and accessing the content digitally. Formed with the December merger of Intel's Interactive Media Services unit and Excalibur Technologies, Convera offers software for managing multimedia content. The vendor's Screening Room software was chosen as the core technology for archiving the training videos.

Before Sandia began researching the project, in August, trainees would have to go to a particular physical location where the tapes were stored and request specific volumes through video librarians. Since actual archiving began in December, 15 of the 500 videos from the 1970s and onward have been archived and transcribed. "We prioritized which ones were [most] important and started with those," says Sandia software engineer Mike Krawczyk. "We were doing it part-time on an ad hoc basis, but we will be doing more full time now."

Krawczyk says the agency chose Convera's Screening Room software over other products because it offered better searching capabilities. "The other systems grabbed the video, but Convera grabbed it and handled the metadata for retrieval," he says.

Screening Room manages video content by examining the underlying binary code and supporting extraction frame by frame. "Screening Room takes the video into the system, digitizes and deconstructs it into component parts, and creates a visual summary of all significant scenes," says Dan Agan, Convera's senior vice president of corporate market development. "You have to determine when a scene change occurs, then capture the scene [metadata] and put it into a storyboard or thumbnail that you can query."

Screening Room can also enhance content with annotations, including notes on particular scenes or attached documents. "For example, you could make available a resume for a subject matter expert," Agan says. "[You simply] drag, drop and display the document in an annotation field so that it becomes accessible with the video."

Video is made searchable through a text capture feature. "This is natural-language [technology] that deploys conceptual techniques to distinguish between the way something is said and the way it is written," says Agan. "The spoken word is converted into electronic text using speech recognition technology from third-party vendors. Then, all material is handed to the content management server. Users can perform intuitive text searches and they can roll back to precise segments [of the video]."

Screening Room converts spoken words to metadata, and the words are indexed in the content server. You can find exact matches or conceptual matches.

"If you were to query the word 'failure,' you would also get instances using words such as 'disaster,'" Agan explains. Like other search engines, the system ranks the confidence of these word relationships so you start with the best search results.

Agan says the speech recognition technology is 80 percent to 85 percent accurate in translating speech to text. "It's getting better now provided you [don't encounter] bizarre jargon with too many acronyms or specialized terms," he says. "The main problem is trying to separate background music and ambient noise from the spoken word before you do the processing."

Every hour of tape generally takes about four hours to transfer. Some videos are classified and are handled on one server while the declassified videos appear on a different server.

According to Krawczyk, the main benefits of the system have been reduced labor and improved access. "The new system has cut our workload in terms of searching for the videos," says Krawczyk. "In the past, you might want to see a section on thermonuclear explosions [and have to search through] 50 minutes of tape to find it. Now, information is provided much faster, with data delivery in seconds. We are also able to use the system in the field and view the segments on laptops for exercises. And we have 100 percent reliability with the text search."

The software cost $75,000, exclusive of consulting services.

"Now Sandia and other government departments can securely provide searchable training information to remote locations, save money by not shipping materials, and view them on their own time on demand, in a personalized and customized manner," Agan says.

In searching for a digital content manager, Krawczyk encourages planners to decide what matters the most. "You have to know if you just want to store the video without the need to search," he explains. "This type of product is in its infancy. We're going to see increasing potential to help out training in various organizations."

U.S. Embassy in Mexico Smoothes Visa Processing

The U.S. Embassy, Mexico City, had a problem, and it showed up every day. The dilemma: 2,000 to 3,000 individuals applying for U.S. visas each day, with the embassy relying on antiquated, paper-based filing and retrieval methods.

To prevent fraudulent entry into the states, embassy officers needed to examine and confirm previously filed information. Lines of people seeking visas were getting longer, because though key information related to each applicant was entered into a master database, the original, handwritten forms and accompanying color photos were stored in filing cabinets. Whenever the documents were needed - say to confirm the identity of a visa applicant against an original photo - it took a minimum of 10 minutes to retrieve the files, and sometimes they couldn't be found at all.

The embassy staff was growing frustrated. "We needed a way to capture and store images and retrieve them rapidly," says Jim Sundstrom, former information services officer at the embassy, "but no one seemed to be able to capture the color [ID photos] in a production environment."

Sundstrom sought out advice from various companies, yet all but one said his mission was impossible. Data Management Internationale, Wilmington, DE, asserted otherwise. The reseller put together a turnkey system incorporating a color-capable 3590C production scanner, from Eastman Kodak, Rochester, NY; Ascent Capture software, from Kofax, Irvine, CA; an Alchemy back-end database, from IMR, Englewood, CO; a storage device, from Raidtec, Alpharetta, GA, and a DVD jukebox for archiving.

The applications and photos are now scanned in, and an operator at an Ascent workstation uses manual indexing to enter the applicant's name, ID number and date. The indexing data is released to the Alchemy database while the scanned images are archived to the DVD library. The system can, in seconds, retrieve applications by name, date or ID number, with photos, handwritten information and signatures instantly available.

"This system was able to handle [a major] problem we encountered," Sundstrom says. "The applications had black-and-white text information on one side and the color ID picture on the other. We needed to capture both simultaneously. The image is always two sided, and documents [from other departments] could be anywhere from two to 20 pages."

The speed of the Kodak scanner helped Sundstrom scale the new system to store more than 2,000 applications per day. "We had to store visa refusals and information from other agencies, such as [those keeping track of people] crossing the border without the proper papers." Sundstrom says the system allowed all related documents to be managed and retrieved from the Alchemy database.

The new system is saving three to five work hours per day at the embassy, says Sundstrom. He adds that it allows users to make corrections in the system while others are online accessing records.

"Mexico is not unlike the United States in that there are a lot of similar names comparable to 'John Smith.' Sometimes we would find someone who appeared to be the same person with the same name and date of birth, and we'd have to check to see if it was a duplicate in the system." Sundstrom says the new system lets agents perform these checks much more quickly than before.

"The issuance of visas is very sensitive politically," says Sundstrom. "We want to be sure that our officers have the correct tools and facilities for this." According to Sundstrom and Data Management Internationale, the technology is now being considered for use in other embassies throughout Latin America.