[Digikam-users] Templates: ITPC subject code

Jean-François Rabasse jean-francois.rabasse at wanadoo.fr
Sat Apr 13 14:53:44 BST 2013


Hello Jürgen,

On Tue, 9 Apr 2013, H. Jürgen Karbach wrote:

>> Controlled vocabulary is a registered fixed standard and
>> the rule is « use it or forget it ».
>
> Oh. I didn't realize, that CV is a fixed standard.
> I believed that it concerns with CV a general name for hierarchical listings
> of concepts of one or several subject ranges; not a monolith.

It is, and for professional usage it's a necessity.
The problem with extensible indexing vocabularies is that when you add
some new keywords/tags, your current database should be browsed again
to identify those images that become relevant, wrt the new tags (tags
that didn't exist at the time the image was archived and indexed).

If you are an amateur, hobyist, with hundreds or thousands of images,
this work remains possible, even though it's work ! More, if you are
the only images provider of your personal collections, your memory
will help looking for ancient images that are to be reindexed with your
new tags.
But if you are a press agency, with tens of millions of images, collected
from hundreds of individuals, photoreporters et al., the a posteriori
reindexing is impossible. So, once in use, the vocabulary must be a fixed
standard.


> E.g. you are working / hobbying on a wedding.
> If you have a managed vocabulary for wedding photos, you can tag them very
> easy by general tags like "Bride", "Bride's father", Bride's mother, Groom,
> ...). So it would be much easier to match the pictures of all togs of a event
> to one homogeneous set of tags.

See what you mean, I happened to do two wedding albums with around 300
images for each one. But I didn't had to face tags inconsistencies issues
because many guests sent me images but no one did the tagging job:-)
(It's very easy to take 10000 pictures of a wedding, with the camera set
on burst mode, but tagging is work and most people are lazy.)

But if you work with serious people, it's not a problem to decide and share
a common tags vocabulary, suited for the occasion.
Anyway, be sure not to go too far in your definitions. When I read your
example, I have the feeling this goes perhaps out of the scope of a tags
activity. Tags or keywords are properties, or attributes, you put on an
entity, be it a picture, a document, other. And yes, « bride » is such
an image property. But what about « brides'father » ? To me it sounds
rather like a relation between two entities. You can say « X is the father
of bride Y », but this makes sense in a relational database, and when one
of your entities, X or Y, is removed, the relation disappears.

But it's another topic and, in my opinion, it's not the kind of job that
could be expected from amateurs images management software. Indexing
databases is a specific task that lies outside images edition issues
and should probably require dedicated software. And this is not something
that interest a majority of users, that's why images management software
are more or less weak regarding metadata issues, tags issues, etc., and
mostly focus on images edition and processing.


> Btw. Maybe it would be a good idea to extend digiKam's face detection with a
> second column called "profession" or "role" or something in that way, so it
> isn't in digiKam's person-database, but tagged as [role]. You already can
> select it by first-time-naming a person. With the "second column you able to
> name and tag in one go.

Cf. supra, here you start designing an individuals data base, not an images
tagging architecture. Role why not, Profession why not, but someone else
will also want the age or birthdate, and users interested in genealogy and
families trees will also want extra information, « X is the nephew of ... »
For me, looks like a Pandora's box :-)


>> It's also possible to « read » a vocabulary tags file and
>> set all of them to a dummy image. Reading metadata from that
>> image will create all tags. I happened to do such thing but
>>  with a limited number of tags, several tens. Not 22000 : -)
>
> Is there any How-To ?
> Can you give me a hint?

I don't think there's any How-To, because every procedure is fully
user dependent. What I mentionned is the possibility of tagging images
using command line tools. E.g. you wish to add a location tag to an
image, exiftool will do :
   exiftool -xmp-digikam:tagslist+=Location/France/paris  IMAGE.JPG

NB: exiv2 will do too, but I don't known the syntax.

So it really depends on what you plan to do. Il you just want to create
and set a new tag, probably the Digikam GUI is Ok for that.
If you plan to index a large bunch of images, with some tags series,
writing a couple of shell scripts using commands as above will make it
easy to process images folders.
Of course, after that, you have to update Digikam by selecting images
and running « Reread metadata ».


Regards,
Jean-François


More information about the Digikam-users mailing list