[Data-modeling] Types for subjects

Faye Harris faye at metaweb.com
Thu Jul 2 19:51:54 UTC 2009


+1. Way too many "foo subjects" types. And the expected type of 
something with a media subject would be a generic media type... 
Something that would resolve DA-445, perhaps?

-- Faye


Jeff Prucher wrote:
> <http://www.freebase.com/view/guid/9202a8c04000641f800000000b912218>
>
> This is a thread in which it is proposed that we collapse all the various
> media subject types (book subject, film subject, etc.) into one big (new)
> Media Subject type.  Functionally, this would result in one type with many
> properties, rather than many types with one property each.
>
> The main advantage I can see to this is that users wouldn't have to add new
> types if they wanted to indicate that a topic was a subject of a type it
> didn't already have (that is, if you wanted to say that a particular radio
> program had a subject of, say, "Porcupines", but the Porcupine topic was
> only the subject of a book currently, you wouldn't have to add the type
> "Radio Subject" -- you could just fill in the property).
>
> A minor advantage _might_ be that people creating types in bases would have
> a type they could all reference if they had a new type that needed a subject
> property. Although this also has the disadvantage that these properties
> would not be reciprocated, whereas if the user created a new "foo subject"
> type, it would be.
>
> This would also be a fairly major change, since we'd be nuking a lot of
> types, which would potentially break all kinds of things (we could keep the
> old property keys, but I'm not sure about the type keys). It also has
> implications for other places where similar patterns exist (media genres and
> ficitional characters are two that come to mind).
>
> What do you think about this? 
>
> Jeff
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
>
>   



More information about the Data-modeling mailing list