Using keywords & catalog sets to organize on subject matter

[Marc Sabatella]

I've seen the topic touched on in a number of threads, of course, but I'd love to see people who have systems they like tell us about them.  I'm not just talking about controlled vocabulary issues, but broader ones.  I am talking about deciding what sorts of things go into keywords, what sorts of things are organized by catalog sets, and what things are organized both ways.  How do people go about this?

So far I'm zeroing in on a catalog set hierarchy that seems like it will work for me, but my keywording is pretty erratic.  I'm a (mostly) non-professional photographer who is really a musician and landscape painter, and not coincidentally I shoot a lot of musicians and landscapes.  My top level catalog sets include:

Art (subsets for artists, paintings, sculpture)
Events (subsets for weddings, parties, specific events)
Jobs (someday I may have more than one of these...)
Music (subsets broken down by genre and by region)
People (subsets for specific individuals and one for candids of unknown people)
Places (subsets broken down by location and by type - landscape, cityscape, night, architecture, etc)
Things (subsets for inanimate objects, plants, and animals, the latter also broken down into cats, dogs, etc)

Plus of course, various "utility" sets that are not subject matter specific and are not really relevant to this particular discussion.

My cataloging software has a searchable "Notes" field that is free text, and I use it for a brief description that applies to a whole batch of pictures ("hiking @ Mount Falcon", "honeymoon @ Venice").  For keywords, I am currently duplicating some but not all of the catalog set information.  For example, if a picture is in the Music/Genre/Jazz and Music/Region/Colorado catalog sets, it gets keywords music and jazz at least.  Rather than duplicating region information, I will often use a keyword indicating the specific venue.  Also a keyword for the specific musician(s) pictured.  But for other types of pictures, I'm a lot more haphazard.  Many of my pictures in the Places/Type/Landscape catalog set have no keywords at all.  Location information is contained in the Notes field.  I suppose I should at least have a keyword of "landscape".  If I were more industrious, I'd add keywords that more specifically listed the elements.  But I don't know, most landscapes have trees, grass, sky, etc - unless it's something really out of the ordinary ("sunset") it doesn't seem worth my time to note anything more specific.  If I painted from photographs more I might care ("gee, I feel like painting a landscape with a grouping of trees and a pond - wonder if I have anything like that?"), but I don't paint from photos, so I don't need to answer those types of questions.

As you can see, my "system" pretty haphazard, and I'd like to improve.  I'm not really concerned about others browsing my images using keywords, so I'm pretty lax about using them at all.  Mostly its either for information I don't want to risk losing in the event of database failure or migration, or for information useful to me but perhaps too fine-grained to be worth using catalog sets.  Like "train" as a keyword in a picture that is in the Places/Type/Cityscape and Places/Location/Cinque_Terre categories.  Pictures in this latter categories also have keywords of "Europe", "Italy" and "Cinque_Terre".  But the same picture taken here in Denver might just get the "train" keyword, and another taken after the train has passed might get no keywords at all.  I'm thinking I should probably take all the major category names ("cityscape", "animal", "wedding", etc) and push them into keywords also.  Do others do this regularly?

Anyhow, I'm just interested to see other people's overviews of their systems, to give me ideas.  And I'm assuming I'm not the only one curious about this...

[danaltick]

Marc,

You've expressed a lot of legitimate concerns.  Concerns that I've faced myself and am still facing today.  However, I think that metalogging is something that takes practice, patience, and continuous learning to get better at.  It must become an integral part of good catalog management, and whether your images are strictly personal or public, I believe it's still a good habit to get into to always attempt good metalogging.  A good place to start is by reading the articles at www.controlledvocabulary.com.  Make sure you read the article on keywording, that is the Who, What, When, Where, Why, and How of keywording.  The IPTC Caption/Description and IPTC keywords fields are the most commonly searched IPTC fields.  I would place strong emphasis on them and follow David's guidelines.  I believe it does also help to either develop a controlled vocabulary of some sort or purchase one.  I use the Image Information Toolkit along with David's keyword catalog.  I will admit, I've gotten somewhat lax with it lately, but after having recently read back through David's articles, I'm going to revisit my star rated images again and do a more thorough job of metalogging, if for anything, just for the practice.  It is my hope that my images may one day become valuable to people other than just myself.  Good metalogging will prepare them for that day.

Dan

[riecks]

Marc:

I think you really need to ask yourself why you are doing this. Is this primarily so you, yourself can find your images, or is it so that others at some point in the future can find images on a particular topic, or of a particular person, place or thing?

I have put most of my efforts into "metalogging" and making sure that keywords and captions are entered into images. I often take a two step approach, putting a minimal amount of keywords for all the images. For those "selects" I put in a great deal more effort.

Part of the reason is that I've been doing this for a long time (starting back when there were only a few choices on the market for image databases), and well before options like "catalog sets" were available.

The downside is that it may take more time up front. The upside is that once it's done and embedded in your primary archive image it will also by default, be in all your derivative versions. In addition, when you send that image out, it's already got metadata in it, and others that drop it into their catalog have the same caption and keyword information as you do. They can then "find" that image by any of the keywords present, or based upon words in the caption.

Provided that this metalog information is stored using IPTC Core, or even IPTC legacy (Image Resource Block) schemas, it is readable in many, many applications. Thus if your images are destined to travel, captions and keywords will serve you better than catalog sets.

Which circles back around to why you are doing this. If the work is primarily for you, then do the minimum necessary so that you can find what you are looking for. It's far better to do a minimal amount of work on nearly everything, than it is to only annotate a few images really well.

David

[Marc Sabatella]

I think you really need to ask yourself why you are doing this. Is this primarily so you, yourself can find your images, or is it so that others at some point in the future can find images on a particular topic, or of a particular person, place or thing?

My intiial reaction to this was "of course, but I'm not that interested in others searching my images, so this answer doesn't help".  But on further reflection, I realized it is still quite useful.  If I'm looking for a meaningful way to decide between using keywords or catalog sets to organize on a given type of information, and I am truly not concerned with others searching my images, then I wouldn't need keywords at all, really, except for the sorts of haphazard "too detailed to be worth a catalog set" things I was already doing.

However, on *further* reflection, I decided, "why not go ahead and get this distinction right *now*, while I'm still actively working on importing legacy images"?  At first, it seemed it would be a lot of work.  But I quickly came to realize that all the work I had already done sorting into category sets made the job of assigning corresponding keywords trivially simple.  So now, most of those same things I had previously described having orgranized with catalog sets have been pushed into keywords, and catalog sets are reserved primary for things that are little interest to anyone but me (eg, the name of the person whose birthday party the pictures come from).  The main place I am still duplicating information between keywords and catalog sets is in areas where I think I have an broad and exhaustive list of categories and I want to be sure to assign all images to one of these.  For instance, I have "places / by location" broken down into landscape / seascape / cityscape / interior / architecture / home / yard / garden.  All images that are basically of "places" should get at leats one of these keywords, and to remind me of what my choices are, it is useful to see them as catalog sets.  Ultimately, of course, this is a job for controlled vocabulary tools, but for now, this works fine.

Anyhow, I'm still interested to hear about other people's systems...