The short version of this lengthy post is that there's really no value in
worrying about how to handle precoordinated strings except for purposes of
busting them up.

The Rube Goldberg style precoordination rules that cause so many headaches
were developed to address challenges brought about by paper card catalogs.
The physicality of paper required a mechanism to ensure a limited number of
cards would file together. Unless you still use a paper catalog, they're as
relevant as spurs are to race car drivers.

The order you see in the MARC record mimics the paper rules exactly (because
MARC was used mostly for card printing for decades) and has also lead to
literally tens of millions of unique subject strings as there are so many
permutations.  As a practical matter, even highly trained librarians cannot
guess how these were put together without going through a substantial
research process.

I hate to dig up stuff written in the 1920's that's rammed down the throats
of first semester library school students. However, in the case at hand,
logic from these works has direct application for purposes of making MARC
data usable.

To summarize, the concept is that subjects can be broken down into aspects
(i.e. facets) with the primary ones time, place, action, material, and
personality -- you can think of this last category as natural groupings of
the type that standardized subdivisions can be applied to such as materials,
animals, corporate entitities, diseases, body parts, etc.

It's much better to think of the facets (time, place, etc) as attributes
rather than occuring in any particular order as this allows interactive and
relatively precise drilling through huge amounts of data. You'll notice that
good search engines effectively do just that.

kyle




One of the challenges for pre-coordinated strings at least as currently
> implemented (that facets evade) is that no order will suit everyone. Which
> of the following is better?
>
> Dwellings $z Australia $x History $y 20th century
> Dwellings $z Indonesia $x Economic aspects
> Dwellings $z Indonesia $x Psychological aspects
> Dwellings $z Indonesia $x Social aspects
> Dwellings $z Ireland $x Economic aspects
> Dwellings $z Ireland $x Psychological aspects
> Dwellings $z Ireland $x Social aspects
> Dwellings $z Japan $x Economic aspects
> Dwellings $z Japan $x Psychological aspects
> Dwellings $z Japan $x Social aspects
>
> OR (mostly current practice)
>
> *Dwellings $z Australia $x History $y 20th century  **Current practice
> Dwellings $x Economic aspects $z Indonesia
> Dwellings $x Economic aspects $z Ireland
> Dwellings $x Economic aspects $z Japan
> *Dwellings $x History $z Australia $y 20th century  **Airlie recommendation
> Dwellings $x Psychological aspects $z Indonesia
> Dwellings $x Psychological aspects $z Ireland
> Dwellings $x Psychological aspects $z Japan
> Dwellings $x Social aspects $z Indonesia
> Dwellings $x Social aspects $z Ireland
> Dwellings $x Social aspects $z Japan
>