Thanks Ian for bringing up this topic, because many people think that tagging is just a matter of throwing whatever comes to their mind. Which of course brings their tagging system to a big mess.

The biggest frustration for the tagger would be when he 'knows' something is in his database and he can't find it (e.g. he used synonyms or didn't tag the right attributes).

I would like to refer to this issue more deeply and give a real world example - 'arranging your scanned documents using a tagging system'.

First let me introduce myself - my name is Shimon Sofer and I am a co founder of '42Tags' - a software for scanning and tagging documents (Windows only for now).

I would like to show how your ideas are implemented in the field of 'scanned document tagging' using '42Tags', and would like to introduce another idea - 'Major/Minor tags'.

In our application we distinguish between two types of tags. The first type - the 'Major tag' and the second type - the 'Minor tag'.

'Major tags' - should be relatively general and you should have the ability to easily control them,
because there shouldn't be much of them (if you are a home/small office user, 42 will be enough...:) ).

'Minor tags' - are those that are not part of the first group.

I would like to demonstrate the idea:

Let's say that you have a few medical documents which deal with 'a blood test for your daughter Jenny, which
was done to verify if she's got pneumonia. The doctor was John Smith".

In this case I suggest that the 'Major tags' will be "Medical" and "Jenny" and the minor tags will be "blood test", "pneumonia", "John Smith".

The division to attributes you talked about is more relevant to 'Major tags' than 'Minor tags'.

How does a 'Major tag' and a 'Minor tag' practically differ?

In 42Tags we added the ability to add a pushable button for every major tag.

Actually there is only one difference between a 'Major tag' and a 'Minor tag' - 'Major tags' have a shortcut button in a specified tag panel.

Regarding the separation to attributes we offer the 'Tag sections' in which the user should add 'Major tags' of the same attributes.

For example - 'Person section' - Each tag is a name of persons
'Places section' - Each tag is a name of place
'Paper type section' - e.g. Contract, Fax, Invoice, etc...

Anyone can define his own attributes for every section.

If the user would like to reach a 'Minor Tag' - he can just start typing, and auto-completion will help him through it.

Another issue not throughly discussed is the dynamic nature of tags.

After tagging 10-100 items you notice that you need to change your tagging scheme.

For example - you live in Paris and all the papers relevant for your apartment are tagged with
'Paris', and 'Apartment'. After a year you leave to another apartment - in Paris!

You may want to distinguish between the collections.

What we offer is making a search on 'Paris' and 'Apartment', and then using a multiple "Tag/Untag" feature in order for example to add "Rue Barute" to the old material.

Maybe I wrote too much, but I really enjoyed reading your post ;-)

Thanks,
Shimon Sofer
http://www.42tags.com/paperless