I believe we may use similar algorithms.* I think of it as loose
   template matching but I like your description better.* I've
   found I can perform better than most people, not as good as
   experts.* It's satisfying.* I know I'll never get to the level
   of the experts from their perspective, but from the perspective
   of a _majority of people_ including myself, it's damned
   impressive. I found this on the "nearest neighbor" search
   pattern list: "Vector approximation files In high dimensional
   spaces, tree indexing structures become useless because an
   increasing percentage of the nodes need to be examined anyway.
   To speed up linear search, a compressed version of the feature
   vectors stored in RAM is used to prefilter the datasets in a
   first run. The final candidates are determined in a second stage
   using the uncompressed data from the disk for distance
   calculation.[7]" Look at all the important things simultaneously
   (I analogize to flash of intuition - or letting the unconscious
   do some of the grunt work)* draw an approximation around them
   which resembles the final result.* Repeat.* This isn't _exactly_
   it of course.* There's some clustering that occurs as well. My
   favorite algorithm that I use, I don't really know the name for
   it.* Maybe you know the name of the existing one. [because I
   doubt there's anything much that's _new_ - just a retranslation
   of it in variant forms: I start from the thing that's sticking
   up or out first.* The thing at the highest dimension.* Then, I
   lower the dimensions. Then I fold along lines of similarity to
   make a mirror.* Then I solve both parts of the mirror by cutting
   triangles and eating each of the triangles. I may have to fold
   many times - but folding and triangle making are the same. [fold
   something - it makes a triangle of sorts] Another is connecting
   all the highest dimensional things together and seeing what
   triangles THEY make, then eating those triangles, breaking down
   further if I must. There's others, but that's two (or maybe
   one?) method that I use. Step one though is true with all
   methods:* Gather as much information as you possibly can
   maximizing the time for information gathering as much as
   possible before categorization.* Early categorization results in
   losing potential connections. Still working on the
   categorization problem.* That's why I'm putting my brain
   online.* I _really hope_ I can find *some tool* someone made
   that can take my data (I'll transform it however the program
   likes it) that can find unseen patterns _for me_.* Yet the
   process is helping me see them myself. I may have the answer
   already, just not fully realized; I found an autotagger that
   uses synonyms; fits right up my alley.. and perhaps I didn't
   load ALL possible synonyms up yet.* Of course this assumes I
   spelled all words correctly but whatever.* I'm still in
   data-gathering mode anyway.