I am working with a programmer on a new product for identification of
objects. We have come up with a great way to narrow the search and we
are wondering if what we have is unique, or just another technique
taught in computer science. How can I find out if it is truely unqiue
and at the same time not give away my idea? I have hired a patent
attorney and he has done some searches but nothing he has found seems
to fit what we are doing. Is there a specific place I can hire
consultants that will work under non disclosure and analyze my idea?
Are there books or internet sites that I can go to and find out if my
idea has been used before?

As mathtalk states, this is a very gray area, however, if it were me,
without more information on the nature of your idea, I would:

1. Do a Google search and a Google Groups search using terms that
describe that describe your process to see if anyone is suggesting it
as a solution to anyone else. This would also be likely to turn up
existing methods related to yours.

2. Go to the websites of products that claim to perform a similar
function as your algorithm. Look at their descriptions and technical
information in order to glean any hint that their process may be
similar (or not similar) to yours.

3. Go to the local computer bookstore and browse through the algorithm
books looking for your method.

4. Look to the classics. If you are mathematically capable of doing
so, look at books like Donald Knuth's The Art of Computer Programming.
He sat down and did the math on many basic programming concepts.

5. Search the periodicals that specialize in algorithms, such as Dr.
Dobbs and the publications of the ACM. Ask your local universities'
librarian for help.

6. Software patent law is extremely squirrelly. It is difficult to
patent an algorithm. Those algorithms that do have patents are
notable, like the LZW compression patent. A negative patent search
does not mean that the method does not currently exist. If you have an
attorney do the search, find an attorney who specializes in software.

7. Treat your idea as if is were a PhD thesis. PhD theses are supposed
be reasonably unique, therefore, if you do a thesis search, you may be
able to find if your algorithm has been explored by a doctoral
candidate. Universities that specialize in software generally publish
what their programs are researching.

8. Your question can be interpreted as involving some type of visual
recognition algorithm. Nobody knows more about this type of technology
than the military (target recognition, etc.). Look into what they use,
and what universities helped them develop it. You may find something
you recognize in your own algorithm.

9. Eventually, a good bet is probably to cultivate a relationship with
a programmer of sufficient age and experience to have been exposed to
the history of search algorithms and simply ask.

Good luck!

- Hammer

I asked if there was a place I could find consultants who could help
me determine if an algorithm I have was unqiue. I said in my question
I had a patent attorney. The answers I got back where focused on how
to get a patent and gave me no information on how to find people
qualified to help me find out if my algorithm was unique.

Hi, goof166-ga:

Since you provide no detail about the "identification" process, I can
only comment at a very high level.  If you truly feel your idea has
commercial potential, the best course might be to consult with a
patent attorney who specializes in software patents.  If your idea is
fairly specific (visual identification using principal components
analysis, for example), then a patent search may be able to answer
your question.  If it is broad idea with lots of
implementation/application possibilities, then consideration of prior
art will play a significant role in narrowing exactly what might
deserve patent protection.  However software is a young and burgeoning
field, and prior art claims notoriously difficult to document or
disprove.

This is a known difficulty with the application of patent law to
computer software.  The prior art in this field is simply not
well-enough organized to permit any systematic search to be done. 
There have been efforts to organize such material, both by those who
oppose the patenting of software on principle and wish to hinder it as
much as possible by maximizing the body of known prior art and by
those (Patent Office researchers) charged with fully validating claims
of novelty (and non-obviousness) inherent in any patent application.

See this article for some leads and perspectives:

[An Introduction to Searching Software Prior Art]
http://www.ipmall.fplc.edu/hosted_resources/bp98/welch.htm

regards, mathtalk-ga

If the technique is truly unique then it will not be understood
therefore a rather simple solution is developed.
If the technique is slightly visual then your are out of your field
already and in for a dollar run for competition therefore you are on
your own in other words there are ways to determine if it be unique
provided you have the big bucks to keep brains from traveling and
mouths from talking.
Who can analyze this? Anyone you choose?

Try falling first then standing up meaning if the idea is unique and
will profit you 1 billion just dream of how much it will cost to
analize(110%).

#Note if the patent dosent cover anything the ideas are largly
incorperated