Sunday, January 15, 2006

IBM proposal to help patent examiners

Working with other tech firms, IBM's proposes a three-pronged approach to help patent examiners learn when a patent application is sought for software that's already in the public domain.

#1. Searchable database

The computer companies are building a searchable database that includes open source software and the date of its creation to assist patent examiners.

#2. Interactive internet service

They are also working with the patent office to create an interactive Internet service that would give people an easy way to look at patent applications so they can alert patent officials when an application covers something already in use.

#3. Automated system to evaluate quality of application

Also, IBM is working with academic researchers in an effort to create an automated system to evaluate the quality of a patent application. This would be an objective, computer-driven system that would point out when a patent application uses terms, for example, that it doesn't define.


"This is an unprecedented effort," said David Kappos, IBM vice president and assistant general counsel for intellectual-property law.

Traditionally, people in the private sector who invent something hire an attorney, file a patent application and then leave the rest to the government. But technology's quickening pace is too much for any one agency to master, Kappos said, and people in the tech community realize they must do their part if the nation's patent system is to survive.

IBM's initiative aims at patentable software and products related to the Internet, Kappos said. As it begins to make a difference in patent application processing later this year and in 2007, he predicts, the use of interested outsiders who are technically able to help patent examiners will spread to areas of intellectual property outside of software.

The "quality" component comes from work of Professor Wagner, separately the person who cited to the first paper by Quillen and Webster on the 97% patent grant rate years after the 97% number and the first paper had been discredited.

Quality criteria may be found here.

One sub-index is the "quantity" of prior art provided by patent applicant (misidentified as patentee).
Look here

The website indicates that the patent quality algorithm will be

publicly available.
There will be a release of the top and bottom 100 patents.

The "first launch" of PQI is scheduled for
January 2007.

This is not unrelated to discussions by
the USPTO about working with the open source community.

Therein, there is text:

“For years now, we have been hearing concerns from the software community about the patent system,” Under Secretary of Commerce for Intellectual Property Jon Dudas commented. “It is important that those in the open source community are joining USPTO to provide resources that are key to examining software-related applications.”

A follow-up meeting on the new initiatives will be held at the USPTO headquarters in Alexandria, Va., on February 16, 2006. The session will be open to the public and devoted to working on details of the three initiatives and the development of additional proposals.

One does note for all the concerns about the Eolas patent, it survived re-examination, without even an amendment.

The 271blog points to another issue with Polk Wagner's quality index. It may fall within the scope of a patent (6,556,992) issued April 29, 2003.

The first claim of the '992 patent:

1. A computer-automated method for rating or ranking patents or other intangible assets comprising:

selecting a first population of patents having a first quality or characteristic;

selecting a second population of patents having a second quality or characteristic that is different from or assumed to be different from said first quality or characteristic;

providing a computer-accessible database of selected patent metrics representative of or describing particular corresponding characteristics of each said patents in said first and second patent populations;

constructing a computer regression model based on said selected patent metrics, said regression model being operable to input said selected patent metrics for each said patent in said first and second patent populations and to output a corresponding rating or ranking that is generally predictive of the presence or absence of said first and/or second quality in said first and second patent populations according to a determined statistical accuracy; and

using said regression model to rate or rank one or more patents in a third patent population by inputting into said regression model selected patent metrics representative of or describing corresponding characteristics of said one or more patents in said third population to be rated or ranked and causing said regression model to output a corresponding rating or ranking based thereon.

The '992 patent claims priority to provisional patent application Ser. No. 60/154,066, filed Sep. 14, 1999.

Given that Wagner's work is alleged to represent work within the open source community, the presence of a patent with claims possibly covering the endeavor would be a major embarrassment to the PTO and to IBM.

from LinuxInsider
The third leg of the patent initiative stool is the Patent Quality Index. This is an initiative that will create a unified, numeric index to assess the quality of patents and patent applications.

Professor R. Polk Wagner of the University of Pennsylvania will direct the initiative with support from IBM and others. It will be an open, public resource for the patent system.

IBM said the index will be constructed with extensive community input, backed by statistical research. IBM hopes it will become a dynamic, evolving tool with broad applicability for inventors, participants in the marketplace and the USPTO.

Important Efforts
IBM said high-quality patents increase certainty around intellectual property rights, reducing contention and freeing resources to focus on innovation.

The three initiatives are open to all who are interested, and broad participation is encouraged by the initiators. In fact, the USPTO has planned a public meeting to further the projects at its offices in Alexandria, Va., on Feb. 16.

OSDL General Counsel Diane Peters said these important efforts among open-source developers, vendors, end users and government to improve patent quality will reduce potential legal threats to open source developers and businesses.

"OSDL is eager to extend its work on open-source legal initiatives to collaborate with IBM, the U.S. Patent Office and the open-source community," Peters said. "This work will further the adoption of, and confidence in, Linux and open-source technologies."

The Legal Question
What about the legal landscape? Will this initiative reduce potential legal threats and stimulate the adoption of open-source technologies as described by OSDL's Peters?

From the perspective of a software developer, Mueller said the risk will be essentially the same as before. Huge numbers of software patents have already been granted, and countless new ones will be granted on an ongoing basis.

"The potential threat of patent litigation doesn't seem to adversely affect the adoption of Linux and other open-source technologies, despite all of the efforts that Microsoft makes," Mueller said. " But once there is the first successful patent suit against users of Linux or another key open-source program, the situation will change fundamentally."

from TopTechNews:

The open patent review program will permit anyone to submit search criteria and to opt in to receive e-mailed updates with links to newly published patent applications. The goal, according to IBM, is to "establish an open collaborative community review within the patenting process to improve the quality of patent examination." Interested parties will be able to send the patent office feedback informing examiners of prior art that they might not have been aware of.

The second initiative will rely on OSDL, IBM, Red Hat, Novell, and VA Software's to develop a system that will store source code in an electronically searchable format. The Open Source Software Prior Art project will seek to "establish open-source software as potential prior art against patent applications." The system will be available both to patent examiners as well as to the public to help reduce the number of patent disagreements and ensure that patents are issued only for new software inventions.

The Patent Quality Index, under the direction of Professor R. Polk Wagner of the University of Pennsylvania, will create a unified, numeric index through which the quality of patents and patent applications can be assessed. It is hoped that extensive community input underpinned by solid statistical research will lead to an index that is a constantly evolving and dynamic tool with broad applicability for inventors, the software industry, and the patent office.

"This is a great example of how the patent office can reach out to the community and how they can help us where we have difficulty getting prior at," John J. Doll, the commissioner for patents, told the New York Times.

Patents are of particular interests to IBM, which ranked first in terms of private-sector patents in 2005. The industry giant received 2,941 patents, roughly 300 fewer than in 2004. Canon came in second with 1,828 patents received.

"IBM believes that patents should be granted only for ideas that embody genuine scientific progress and technological innovation," said John E. Kelly III, IBM's senior vice president of technology and intellectual property. "Raising the quality of patents will encourage continued investment in research and development by individual inventors, small businesses, corporations, and academic institutions while helping to prevent over-protection that works against innovation and the public interest."

Of the "patent quality index," complaints about the declining quality of patents in the United States were a factor in recent calls for patent reform, such as the recent H.R. 2795. Some of this originated in work by Quillen and Webster, suggesting that the US Patent Office granted 97% of all applications, which work was cited by Professor Wagner in 2005. However, the 97% number was modified by Quillen and Webster themselves and the underlying methodology criticized by a number of authors. In fact, the number is much lower, and invalidity determinations by the courts have been going down, not up, in recent years, suggesting the absence of a systematic increase in the grant of bad patents.

As a point of irony, US 6,556,992 is directed to a computer-automated method for rating or ranking patents including constructing a computer regression model based on selected patent metrics. Using regression techniques on old data, whether in Wagner's approach of that of the '992, may not be the right way to predict the future. Further, the metrics proposed by Wagner would indicate that the Eolas patent is a high quality patent.


Post a Comment

<< Home