[Oberon] filesystem with directories ?

[Douglas G. Danforth]

Greg Haynes wrote:
> I think a system for filing photographs is another good test for
> a system for organizing files. With a digital camera you
> can get so many photos that it is almost impossible to
> find things in conventional structures.  If you have used
> Picasa (from Google, I think) it has an attribute capability
> that lets you group photos using arbitrary numbers of
> tags such as date, country, person, etc.  It
> still takes some time to get things organized, but I
> don't know any way to get this capability just by
> using a conventional directory structure. I guess this is
> more of a database capability than a file system.
> -Greg
What this whole thread comes down to is information retrieval and 
pattern matching.
An associative memory such as that developed by Pentti Kanerva (Sparse 
Distributed
Memory) retrieves all 'records' that partially match your search 
query.   Unfortunately
it is difficult with current technology to implement such memories for 
the input space
is thousands of bits (N).  The memory, itself is a sparse representation 
of all possible
2^N locations but this is very adequate for matching images, doing 
speech recognition,
handwriting recognition, weather prediction, and a host of other tasks.  
NASA was
interested enough in such systems to fund a group of 5 researchers for 
four years.
Stanford built a hardware address decoder for 256 bits (at least one Masters
degree resulted from that).

Humans organize information partially based on emotions (good, bad, 
friendly,
hostile, sweet, sour, etc).

Good information retrieval is not a trivial problem.

-Doug Danforth