[Corpora-List] Patent application for "Referent tracking of portions of reality"

John F. Sowa sowa
Tue Aug 31 14:10:27 CEST 2010


On 8/31/2010 3:25 AM, Alan Rector wrote:
> Have you brought this to the attention of W3C?
> Any similar organisation that might have the clout to fight it.

Do you or anybody else know a simple way to do that?

On 8/31/2010 3:41 AM, Laurence Anthony wrote:
> I haven't read the patent details but I would expect that they use
> some niche technology that hasn't been reported before. In that sense,
> the patent application would be valid. In fact, there are many, many
> patented natural language processing systems that are all quite
> similar. Patent writing is an art and the whole world of patent law is
> very fuzzy and complex.

Those statements are mostly true, but there is a question about the word 'valid'. It is common for patent lawyers to describe an invention and the claims in terminology that is deliberately "obfuscated" to confuse the patent officer who reviews it. They frequently succeed because the patent office is overloaded with applications, and POs can't have adequate training in every aspect of every field for which applications are submitted.

For patents of physical objects and processes, the reviewer's task is often easier. The terminology for describing an object is less significant than the drawings, and there is standard terminology for the chemical elements and compounds.

But you can change the names of all the variables and rewrite all the comments in a program, and every variation will compute exactly the same results. If somebody is granted a software patent, they can threaten to sue anybody who uses an algorithm that is provably equivalent.

Prior publication is sufficient protection. If you publish something prior to the date of a patent application, anything that is described in the publication is protected. But it's important to describe your software with the same terminology used in the publications. That makes it easier to defend against some patent that uses weird words.


> But, I think we can relax and continue working in corpus linguistics
> without worrying too much about it. It only becomes important when
> you take a new technology to a company and try to sell or license it.

That may be true, but researchers need funding for frivolous things like food, shelter, and clothing. Sources that provide funding might be less willing to support projects that use technology that is threatened by patents.

John



More information about the Corpora mailing list