Index vs. Search? (take III)

[Ned Bedinger]

Geoff Hart wrote:
> Jay Maechtlen noted: <<ideally, of course, the search engine would  
> look at synonyms and offer alternative search terms...>>
> 
> Definitely, but creating such a compendium of synonyms would be an  
> awfully thorny challenge in artificial language -- one that isn't  
> even close to having been solved yet. As your example (below)  

Geoff, think cross-references. The synonyms needed are provided in the 
putative useful index.

> indicates, it's not trivial to embody all the implicit knowledge we  
> accumulate over decades of linguistic weirdness. And the resulting  

The thesaurus function is basically trivial--just look it up a word in a 
modern electronic, updated equivalent of Roget or some such.

And I'll bet it doesn't take a highlyskilled software developer to 
extend online help to include synonyms in a Help lookup.

On second thought, I don't want to bet on that. Maybe it would require 
someone who works in the Help system's API, to hack in a call to an 
add-on synonym retrieval function that passes the synonyms back to the 
search function. But it's not a show stopper of a problem.

> database of synonyms would be huge, even if you constrain it heavily  
> towards a specific domain (e.g., word processing).

Yeah, it is "huge".  But so is the power of a database lookup function. 
  And while it is shocking to realize that the price of bandwidth and 
workstation hardware today is low enough that every workstation could 
have a thesaurus onboard, a look around will reveal that a modest 
investment of time and cash can make anyone able to download several 
gigabytes of thesaurus data onto a thumb drive (or workstation hard disk 
or gigabyte network drive), and crunch it with a ferocious CPU and 
database combination that will perform the sucking out of synonym search 
results, from those gigs of data, in the blink of an eye.  We're there, 
man!  But it ain't AI--this is a baby step, where the user provides the 
intelligence, and the computer(s) execute their mundane search and 
retrieval stuff.

HTH,

Ned Bedinger
doc at edwordsmith.com