Although I will reply to Tom's questions soon, what suggestions do you have to ensure that I'm not optimizing incorrectly?

----- Original Message -----
From: Geoff Bowers <[EMAIL PROTECTED]>
Date: Thursday, December 11, 2003 3:15 pm
Subject: Re: How to make ColdFusion Verity/search better?

> Tom,
>
> Tom Jordahl wrote:
> > Sample questions:
> > - Do you use our Search?  Another technology?  Why? Why not?
>
> We use Verity search quite a bit.  Why?  Well it ships with CF and
> its
> dead simple to set up. The only times we don't use it are when the
> client already has an existing search technology they would prefer
> to
> leverage.  Or perhaps if we need to spider multiple sites and
> provide an
> integrated collection.
>
> > - Do you use a K2 server?  Did you know it goes faster if you
> do? :-)
>
> Sometimes we use K2 if the search is actually been hit quite a
> bit.
> Generally we don't bother -- i think this is a bit of a legacy
> "mind-set" issue in that you couldn't simply use the same
> collection
> name for both VDK and K2 -- K2 required a unique alias name.  And
> you
> had to start and configure an additional service to run on the
> server.
> Not all that hard but I gave up explaining it to folks over and over.
>
> I can't remember if the aliasing issue was resolved in CFMX.
>
> > - What document types do you index?
>
> Mainly database queries and PDF.  Occaisionally we'll be involved
> in an
> intranet project where all sorts of file types are indexed.
>
> > - How many documents do you index?
>
> Varies greatly.  As a side note, I'm not sure that individual
> records in
> a DB query should be treated as individual documents.  Why not do
> this
> licensing on the basis of collection size?
>
> > - How many collections do you have?
>
> Our approach has generally been to have a separate collection per
> document or content type.  Makes it easier to offer advanced
> search
> options and filtering.
>
> > - What features do you like?
>
> I really like that in CF all Verity collections look the same.  So
> it
> doesn't matter if you are working with PDF's or database queries
> or a
> combination you can represent the results all in a single query.
>
> > - What features are we missing?
>
> -- maintenance
> Maybe this is changed in later versions of Verity... but you
> really need
> a better explanation of when to use CFINDEX actions, in particular
> UPDATE vs REFRESH.  An explanation of why and how often to
> OPTIMISE and
> so on.  So many folks give up on Verity because they update it so
> inefficiently, don't optimise and wonder why it runs slowly.  
> Actually
> at that point most folk just say Verity "sucks hard" and cease
> wondering.
> --spider
> I find this spider thing a nightmare.  Not really an issue with
> the
> technology more an issue with the licensing.  There are so many
> spiders
> out there... why not make the Verity spider open beyond localhost?
> It's
> not like it offers any huge competitive advantage -- I realise
> this is
> more Verity than Macromedia.
>
> --mapping alternative properties
> I often find myself hacking the collection style files to include
> things
> like datelastupdated in the query output.  Perhaps for files at
> least
> this should be there by default.  What about file size, create
> date and
> other metadata??  Much of this is already in there, just not
> exposed by
> default in the CF/Verity gateway.
>
> --additional search types
> Fuzzy logic, synonym, phonetic and the like.  I'd like to have the
> option of setting up my own lexicon of terms to map to synonyms
> and the
> like.  These are all options in Verity Enterprise... so I'm not
> sure
> what is on offer in their v5 SDK.
>
> That's all I can think of for now.
>
> -- geoff
> http://www.daemon.com.au/
>
>
>
>
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings]

Reply via email to