[Data-modeling] Suggested changes to the military conflict schema

[Bryan Cheung]

Just sending out a feeler to see if there is any feedback to Jeff's  
suggested changes; otherwise, I'd like to go forward with his  
proposal...

Bryan


On Mar 10, 2008, at 3:39 PM, Jeff Prucher wrote:

> This is a task from months ago that finally bubbled up to the top of  
> my
> queue.  We had a request in the military discussion posts requesting a
> better way to model the relationships between actors in a military  
> conflict.
> The problem is how to represent groups of combatants fighting on the  
> same
> side in a conflict (a conflict, in this schema, can be anything in  
> scale
> from a single battle to an entire war). The current model represents  
> this as
> a standard type (i.e., one that has a name), which requires that  
> each battle
> have two combatants with names like "Tet Offensive anticommunist  
> forces" to
> fully represent which forces (countries, divisions, units, etc.)  
> fought on a
> given side. (See
> http://www.freebase.com/view/guid/9202a8c04000641f8000000004f3cbee  
> for an
> example.)  This a) is a pain to use, b) is sort of ugly, c) makes  
> automated
> data-importing very difficult, and d) means that you have to make an  
> extra
> click to see which forces actually fought in the conflict.
>
> My proposal is to make this into a CVT, and allow multiple values  
> for the
> combatants. This has two advantages: first, it avoids the extra step  
> of
> having to create an intermediate type; and second, it means that the
> combatants are all listed on the topic for the conflict itself. A  
> few other
> changes fall out of this, such as how commanders, force strengths, and
> casualties are modeled, but these are all secondary to the main  
> change, and
> I won't go into gory detail here. If you're interested, though,  
> please take
> a look at the schema in my private domain. Please bear in mind that  
> the
> types in my domain are sketches, and I only included properties that  
> would
> be affected. Other properties in the current schema will be kept.  
> Please
> also note that the names of my types are not serious suggestions --  
> I just
> wanted to avoid confusion with the current military types; in most  
> cases
> we'd keep the existing types.
>
> Schema: http://www.freebase.com/view/domain/user/jeff/default_domain
> Example data: http://www.freebase.com/view/en/colmar_pocket
>
> Jeff Prucher
>
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling