[Digikam-users] tag structure

[David Talmage]

On Tue Jun 22  9:32 , Benedikt Rudolph <benedikt.rudolph at gmail.com> sent:

> ...
>But now I'd like to have a tag "park" which is a super-tag of "English Garden"
and "Hyde Park". Thus, I'd be in need of a tag structure like /park/English
Garden and /park/Hyde Park. However, as far as I know I cannot assign both,
/city/Munich and /park, as super-tags to English Garden.
>
>
>Of course it would be possible to create to distinct tag structures
/city/Munich/English Garden and /park/English Garden. I would then have to tag
every picture twice with two distinct "English Garden"-tags. This is quite ugly
in my opinion for two reasons: First, there is a lot of redundant information
stored in the image meta-data. Second, the semantics (i.e. the English Garden is
a park and the English Garden is located in Munich) cannot be expressed and
changed (!) without touching each single image.

This is related to the ontology vs. tagging problem that Clay Shirky wrote about
it http://www.shirky.com/writings/ontology_overrated.html.  It's an interesting
article if you have the time to read it.  He talks about Yahoo's ontology vs.
del.icio.us's tags.

This excerpt from the article shows some of the characteristics of the subject
domains and user communities that are best served by ontologies:

[Begin excerpt]
Domain to be Organized

    * Small corpus
    * Formal categories
    * Stable entities
    * Restricted entities
    * Clear edges 

...

Participants

    * Expert catalogers
    * Authoritative source of judgment
    * Coordinated users
    * Expert users 
[End excerpt]



This excerpt from the article shows some of the characteristics of the subject
domains and user communities that are best served by tags:

[Begin excerpt]
Domain

    * Large corpus
    * No formal categories
    * Unstable entities
    * Unrestricted entities
    * No clear edges 

Participants

    * Uncoordinated users
    * Amateur users
    * Naive catalogers
    * No Authority
[End excerpt]


You might create several ontologies for your pictures but that can lead to
duplication as you observed.  Unless you don't mind that duplication, that
suggests using something less rigid.

> ...
>Question 3.) Do you consider it feasible to use a flat tagging structure and
thereby completely exclude the semantics problem from digiKam?

You could use some of each.  I have a few disjoint ontologies plus a number of
singleton tags.

In my tags, I have a hierarchy for places because a place can't be in two places
at once.  For instance, I have pictures from a trip to Europe last year tagged
with "Europe/Germany/Koln" and "Europe/Switzerland/Valais/Breiten ob Morel". This
lets me select all of the pictures from a continent, country, region, or city. 
The other tags are less rigid, such as "Train" for pictures of trains,
"People/Suzanne" for pictures with my then girlfriend in them, and "Vacation" for
pictures I took on vacation.

The "People" ontology doesn't work well when someone changes her name. 
Fortunately, that doesn't happen very often but when it does, I usually leave the
old tags in place for historical reasons.

David Talmage