VoEvent classification

[Rob Seaman]

Brian wrote:

> 	Thinking in terms of so..I don't think unknown phenomena are
> 	a problem, in fact this is where so tech can really shine.

Yes.

> 	Pesumbably you will be describing these events using already
> 	known set of terms/definitions/properties ("luminosity",
> 	"sky location", various types of variablity, etc) which may be
> 	used to infer what the event likely is, so an initial classification
> 	is possible by the machine in terms of known vocabulary.

(Hey!  "Pesumbably" is a Googlewhack!  It asks you if you meant  
"presumbably" - 16,400 hits.)

"The Hunt for Red October" had an example of what you describe.  The  
Russian sub had a new type of silent magneto-hydrodynamic  
propulsion.  The American sonar computer classified this as a "magma  
displacement".

Which is to say that ultimately a human has to propose a physically  
coherent explanation.

> 	Of course, contemporanteous and other simultaneous observations
> 	are often needed to be sure what something _really_ is, and
> 	much later perhaps that kind of information can be made part
> 	of the system so that the machine refines its classification as
> 	more inputs are available.

That summarizes the point of VOEvent nicely.

- Rob