R&D pacts to supportsearch andretrieval

The Patent and Trademark Office in conjunction with the Defense Advanced Research Projects Agency (DARPA) recently awarded its firstever research and development contracts to help administer the complex patent process. Under the agreements which total $11 million the University of Massachusetts a

The Patent and Trademark Office in conjunction with the Defense Advanced Research Proj-ects Agency (DARPA) recently awarded its first-ever research and development contracts to help administer the complex patent process.

Under the agreements which total $11 million the University of Massachusetts and the National Science Foundation's San Diego Supercomputer Center will develop search and retrieval methods that PTO officials can use to manage patent images and text. The new methods will support the automation programs PTO is developing.

The research and development effort was mandated last year by the House Appropriations Committee which sanctioned the use of PTO funds for basic research to facilitate the search and distribution of patent information.

"This is truly a research and development endeavor. This is not normally the business of PTO " said Dennis Shaw PTO's chief information officer. "Most agencies are precluded from doing work such as this and we would be too if we hadn't been directed by the Congress to do it."

The contractors also will test and evaluate electronic filing and security applications that could be used agencywide as the government moves toward a comprehensive electronic commerce system.

But much of the work included under the two contracts will focus on PTO's internal needs.

Within three to four years the methods developed under the contracts will be accessible by patent examiners who now wrestle with the cumbersome job of sorting patents. Through the Automated Patent System contractor PRC Inc. automated many of those functions. But the agency's 2 000 examiners are still stymied in their efforts because applications are not fully deployed to the desktop.

"We made a decision almost a year ago to migrate to Windows NT at the workstation " Shaw said. "Right now we have clusters of computers that our examiners have to walk to" to access the agency's huge database of patent information which includes an image database of 6 terabytes and a text database of about 135G. Through a $25 million hardware buy due for award this fall PTO will save its examiners the trip by providing up to 13 880 desktop PCs.

The R&D work will be done in tandem with the migration said Bruce Lehman assistant secretary of Commerce and commissioner of PTO. "One of the nice things about these contracts is that we are addressing sticky or unresolved problems&hellip before we migrate toward a fully electronic system."

Under the new system companies will file electronically for patents. "It will be a paperless office but our initial focus is on the examination process " Lehman said. Patents can be complex combinations of images and text. Searching retrieving and matching those files has long been a cumbersome process for the agency. "We have all of the patent drawings submitted to the agency since 1790 " Lehman added.

In the past examiners had to manually sift through patents establishing classifications of items. The rapid technology advance has made that process increasingly difficult even with the PTO modernization of the 1980s. The R&D contracts will take on the most sophisticated of the agency's remaining problems. For example the University of Massachusetts will focus its efforts on natural language developing methods that assist examiners in making connections between patents.

"You could call it artificial intelligence but really it is natural language queries that will help with classifications " Shaw said. "The other UMass effort involves pattern recognition which could be of assistance in designs. Forty percent of all patents have designs tied to them."

"We are creating a test-bed system for handling complex documents " said Dan Bender project administrator at the San Diego Supercomputer Center. The center has recently been working on archival and storage methods as well as computational systems on high-performance networks in a "heterogeneous environment " he said.

The center also will study intelligent document management systems. "This will lay the groundwork for an information technology structure once we start receiving patent applications through electronic commerce " Shaw said.

DARPA has retained a strong role in managing the contracts and in providing networking resources necessary to develop the applications according to Larry Cogut PTO's project manager. "PTO and DARPA are really strong partners in this " he said.