controlled vocabulary with iView catalog sets?

[andris]

Is there a good way to implement controlled vocabulary with catalog sets?  I like the idea of purchasing a keyword list and sticking to that list, but I agree with general consensus here that in IVMP catalog sets are a more accessible place to store keyword information due to their ability to work as hierarchical rather than flat.  I'm working on building my own hierarchical list in catalog sets of things like ethnicities, genders, ages, concepts, props, and so on.  I can't imagine trying to apply all of this information if I had to hunt through a giant flat list.  I understand that the cvkc (controlledvocabulary.com) is a list of hierarchical keywords, but iView doesn't display nested keywords.

If I plunk down the $65 at controlledvocabulary.com, how would I get the list into catalog sets?  Assign all 11,000 keywords, and then run a script?  Use the Bridge CS3 file info panel along with some creative finding and replacing of ;'s and |'s?  The photographer I work for is thinking about setting up her own stock site, and all of this searchable keywording is going to become much important for us in the near future. 

Thanks for the help,

Andris

[danaltick]

Andris,

You may want to rethink your current view on hierarchies.  I tend to agree with John Beardy's view http://thedambook.com/smf/index.php?topic=2625.msg14831#msg14831.

Dan

[andris]

Hmm, certainly some good points there.  Still, it seems like a flat list in iView might make the keyword application process more difficult.  In our studio, we have a stack of full hard drives dating back to 2004 or so with little or no metadata applied.  We have an intern who comes in twice a week to do some of the keywording for us, and with the current hierarchy I can tell him to make sure that each image gets assigned to a subset from the gender set (male, female, or both), gets assigned to at least one subset of the ethnicity set (or more for group shots), gets assigned to at least one subset of the age set (or more for group shots) and so on.  Each group of images he catalogs, he goes through the hierarchy and has a clear list of what kind of information he needs to add. 

If I gave him a flat list, and a series of images how would he know what to add?  Would he just start typing in keywords as they came to mind?  Am I missing something?

Also, I definitely see how a flat list and a direct keyword search for the bottom level word makes sense for something concrete, like a moose.  However, when dealing with more abstract concepts it feels like the flat list breaks down a bit.  For example, a lot of stock agencies use a field called 'concept' which is supposed to define the overall 'feel' of the image.  The field contains terms like "happiness, mischief, love, bizarre."  If words as widely varying as those weren't unified under a common heading, I think the idea of 'concept' is kind of lost.

Thanks,

Andris

[danaltick]

Andris,

Keywording is really a skill you develop over time as you gain experience.  I will admit, I'm not that experienced at it, but I do believe it should take priority over catalog sets.  Once my images are metalogged sufficiently, only then would I start grouping them.  I would group them as I need to use the groups (e.g. portfolio searches, client/project, stock, slideshows, personal, job, etc.).  Just because keywords cannot be displayed hierarchically by most catalog managers, doesn't mean you shouldn't still keyword them that way.  As John points out you can actually do searches faster in iView than you can traverse nested catalog sets.  I use the controlled vocabularly keyword catalog whenever I need very thorough keywording of my images, but I really on go to that extent for images I would be submitting for stock or possibly delivering to a client.  To do that to my entire archive would be very time consuming with very little return on the investment.  It would also slow my workflow considerably.  I've found that a half dozen or so good relevant keywords usually is good enough for me to find any image quickly in my archive should the need arise.  I love to shoot, and I don't like something like keywording deterring me from shooting as much as I would like.  What you mentioned above can be done just as easily with keywords.  Instead of spending that time creating sets for your intern, you would be better served spending it doing some top level preliminary keywording for him, then each day do a quick search on one of the keywords and create a quick catalog set from that search.  You've now grouped the images for him to work on for that day and created the set as you needed it.... make sense?

Dan

[andris]

I agree with you that the fast searchability of keywords makes them attractive, after all...Google didn't get where it is today by providing a detailed nested list of all the sites on the internet.   I also agree that it's important to avoid overkeywording...this can be very time consuming.  I don't think I follow you on how you'd use keywords to guide the intern's keywording process, though.  I wasn't implying that I was creating catalog sets to tell him which images to work on, rather the catalog sets tell him which concepts need to be addressed.  I set him to work on a shoot's worth of images (generally 4000 or so), that were already divided into subfolders for each of the day's shots at capture time.  Generally images in each shot subfolder are very similar, and can almost be treated as a single image.

The catalog sets I have him using look something like:

Content Keywords
 |->Age
     |->Infant
     |->Toddler
     |->Child
     |->Teen
     |->Adult
     |->Senior
 |->Race
     |->White
     |->Pan-Asian
     |->Black
 |->Sex
     |->Male
     |->Female

...and so on.  Generally I tell him to start at the top of the list of 'content keyword' catalog sets, and to see which words apply to the images from the shot he's looking at.  He usually does a 'select all' on an entire shot's images, right clicks on a keyword set (i.e. 'infant') and selects 'add term to selected media items.'   With this in front of him, it's obvious that each image of people should be tagged with the ages, races, and sexes of the people in the frame.  And they can be added in a few quick clicks.

Perhaps what I need to do is to leave things like age and race that are objective and contain a relatively small number of subsets as catalog sets and should start using controlled vocabulary keywording for things like 'concept' that are very subjective and who's number of subsets would tend to bloat over time.  I think writing this post has just helped me come up with a new strategy for this.  Thanks!

Andris

[danaltick]

Andris,

I believe I would rather use those content keywords as keywords and not catalog sets.  In otherwords, just give him that list of keywords and let him use that list to keyword the images rather than dragging them to the catalog sets.  In otherwords, I don't see any real need to create those catalog sets unless you need to frequently view all your Seniors, Adults, Males, etc.

Dan

[peterkrogh]

Andris,
I think that Catalog Sets is the place to do this in iView.  Beardsworth's translator would let you make the catalog sets into Lightroom or Bridge keywords as a batch process.

Do you really need to assign the whole collection a bunch of arbitrary terms as keywords, or is the more cost-effective thing to do some basic grouping (portraits, still lifes, client name, project name, etc) first.

Once you get down to a smallish number, then running through a keywording routine would make more sense to me.

It would be interesting to talk to David about how this could be converted to iView Catalog Sets. If you could buy a dummy catalog with all the sets in it, you could merge that catalog with an existing one, and all the catalog sets would show up.

Running the images through Beardsworths script could translate the Catalog Sets into Hierarchical Keywords.

Peter

[andris]

Hi Peter,

I think I probably should have been more specific about our collection at the outset.  Our photographer shoots lifestyle for advertising and editorial clients.  Probably something like 98% of the images in the collection are of happy people doing fun things. 

I think for our collection, basic people descriptions _are_ our basic groupings.  Things like portraits and still lifes do make occasional appearances...but they're very rare.  The broadest division is probably kids vs adults, which can be refined a bit by age.  Since demographics are so important in our market, the race division is another broad and useful one.  This would allow us to, say, find all the 3 star images with pan-asians to pull images for a tailored portfolio for an ad job we're up for.  Other good candidates for short list people descriptors might be group size (single, couple, small group, large group), sex (male, female), and so on.

...that takes care of the 'happy people,' it's the 'fun things' that are harder to quantify.  We have a catalog set of activities that contains things like 'cleaning, cooking, dancing, jumping,' but that one seems to bloat a bit.  Our 'concepts' set is another one that's tough to pin to a short list.  I'm thinking both of those things might be better as keywords with a controlled vocabulary?  Then I could just tell the intern to type the first 3-5 keywords that come to mind into the keyword field and we'd instantly get all the synonyms.  We do frequently get portfolio requests like 'have you shot anything with kids washing a car.'  My search would be two part...narrow by the age catalog set, and then run a keyword search for'car wash'.

I'm still thinking a split between catalog sets and keywords might be the best for this.  I can see myself typing 'basketball' into a keyword search...but somehow searching for something like age, race, or sex where there are a very finite number of options seems more natural as a catalog set.  Conversely, adding things to catalog sets is easy when the list of choices is short....but typing a specific activity (say dancing) seems easier than scanning through a huge list of possible activities.

An alternate way to sell/distribute the controlled vocabulary as iView catalog sets might be to distribute a small set of 'carrier images' into which the whole vocabulary had been pasted via the Bridge file info panel.

[danaltick]

Andris,

I believe I understand your business more now.  I think you're on the right track, and that is, try to define what your basic groupings are, and make those your catalog sets, attempting to keep them to a minimum, then follow that up with a good controlled vocabulary of keywords.

Dan