14

I am using Digikam for photo management, Gallery3 as a web gallery and many other tools offering "tags" as a means of organisation. Though very simple in usage tags offer a great and intuitive way to narrow down search results. On first sight the usefulness of tags depends on two major facts:
1.) usage of tags throught whole collections
2.) senseful selection of tags
3.) reuse of existing tags instead of synonyms
There is one central problem within the system: in larger collections, especially when you share collections, the set of tags becomes to big to be useful. You loose overview what tags actually exist, which ones you want to pick. Too many similar tags come into use, so for example instead of looking for a 'house' you also have to look for 'building', 'home', 'hut', ...
I found I get extremely good results trying to prevent this problem by using tag hierarchies. You still can define tags with a few key strokes, but you have to specify less tags for each item. Especially you dont have to specify "wider" tags like for example 'nature', 'sea', 'landscape' when you tag something as 'costline'. As a benefit you can now even select by these hierarchies: select all 'sea' tags or only the specialized 'costline' ones. It is unbelievable how fast you can pick items from huge collections that way.
That is, if you have a hierarchy. A good one. A logical one. A growing one.
This issue becomes much more pressing when you share items and their tags, since combining the tag sets only makes much sense when the two hierarchies match. Otherwise things get annoying.

Why don't we start and create a standard for tag hierarchies ? Something like the Dublin Core for Tags ?
Still everyone is free to define own additional tags or even ignore the standard if he chooses to. All fine.
But it allows to get faster and better working results when building up a tag database. You don't have to make all those mistake again, many many others already did.
And your tag hierarchy becomes compatible when sharing collections.
And, it might also become possible to get 'multilingual' tags without much effort, since the tags in use can be matched against a standard holding translations for tags.

I imagine a central database (say on freedesktop.org) with an API and a simple web frontend. Applications can either import a hierarchy or parts of it for local use using that API. Or they could check new tags against that database and make suggestions how to write a tag and where to place it in the hierarchy.

OpenSource and Collaboration in projects is all about re-use of existing things. So it would certainly make sense not to reinvent the wheel again and define a new hierarchical structure of expressions, but to look out for such a definition. I guess such structures already exist somewhere in the fields of linguistics or logic. If so it certainly would make sense to take those as a starting point.

Does anyone know about such hierarchy definitions ?
Is there already something like an "expression relationship diagram" defined ?
Not that such hierarchy does not necessarily have to be a strict tree, it might very well contain loops and cross links.

Ok, been a while, but it appears that slowly the wheels start to turn, still with a grinding noise, but a movement at last !
There is a small group of people who pick up an exchange of a few ideas and aspects that come into play, when thinking on with this idea. And since KDE is all about community and getting involved I want to confirm my invitation that this idea really needs supporters. Any input is welcome, and I certainly could imagine an exchange with people able to contribute to a prototype. Especially the GUI turns out to be much more challenging than expected, an intuitive and fast interface is extremely important for the acceptance of the idea. So drop me a note ! Give me hints ! Contribute !
---
arkascha