[Data-modeling] Dates as Events (was Astronomy: Discovery Properties)

[Robert Cook]

We have gone through this several times now.  Using an abstract event  
type makes certain queries easier at the expense of complicating every  
data model that it touches.  Given that the most important thing right  
now is to acquire data so that these expected queries meaningful, it  
seems a poor tradeoff.

As use cases come along that require more unified semantics on time  
based properties, we can add structures to the graph to support them  
-- structures on top of properties that describe common meaning (since  
we have a graph, we can do things like this.)  Right now it's not at  
all clear to me that the very generalized event type that has been  
proposed will capture the subtleties of time in these various contexts  
-- in other words it is most likely a premature model optimization.

R
On May 6, 2008, at 1:25 PM, Jeff Prucher wrote:

> I'm pulling this discussion out of the astronomy thread since it  
> touches on
> things that people who might have stopped reading this thread might  
> want to
> weigh in on.
>
> Didn't we go down this road awhile back when we first started  
> talking about
> an events type?  It seems ever-so-much simpler to put a property  
> with an
> expected type of date/time for this sort of value. Otherwise, we get  
> into
> the sort of modeling where every date assertion should actually be  
> an event,
> and we end up with a model where every person has an associated  
> event "birth
> of X", and all deceased people have a "death of X" event, etc. ad  
> infinitum.
> A data-driven timeline like you suggest can already be made by  
> selecting the
> date/time properties from the types you're interested in, without a
> proliferation of topics like "Jeff Prucher graduates from college".
>
> Jeff Prucher
>
>
>> -----Original Message-----
>> From: data-modeling-bounces at freebase.com
>> [mailto:data-modeling-bounces at freebase.com] On Behalf Of Danny Hillis
>> Sent: Monday, May 05, 2008 6:45 PM
>> To: Freebase data modeling mailing list
>> Subject: Re: [Data-modeling] Astronomy: Discovery Properties
>>
>> I was trying to imagine a query that would access the
>> "discovery date", and was imagining the use case of plotting
>> a timeline of discoveries, inventions, etc. in the context of
>> other historical events.  If there were topics for historical
>> events, then content of the timeline could be data driven,
>> selected by the categories or genres of the topics.
>> Otherwise, I don't see how to do it without an ad-hoc query
>> for each type discovery, invention, etc.
>> -Danny
>>
>> On May 5, 2008, at 4:35 PM, Jeff Prucher wrote:
>>
>>> To model it as an event, assuming that this means to use
>> the "event"
>>> type to
>>> record information about the discovery of an object, requires the
>>> creation of topics like "The Discovery of Pluto". But all using the
>>> "event"
>>> type gets
>>> us is the start and end dates, and an inclusion/included by
>> tree, the
>>> latter of which probably doesn't apply. This actually seems
>> like more
>>> work for less payoff than just using the properties "discovered by"
>>> and "discovery date", since "discovered by" would have to
>> be on a new
>>> type ("astronomical
>>> discovery") that included the type "event". Unless you had
>> something
>>> else in mind, in which case kindly ignore this message.
>>>
>>> Jeff
>>>
>>>
>>>> -----Original Message-----
>>>> From: data-modeling-bounces at freebase.com
>>>> [mailto:data-modeling-bounces at freebase.com] On Behalf Of Gordon
>>>> Mackenzie
>>>> Sent: Monday, May 05, 2008 4:00 PM
>>>> To: Freebase data modeling mailing list
>>>> Subject: Re: [Data-modeling] Astronomy: Discovery Properties
>>>>
>>>> No I haven't really considered it as an event.
>>>>
>>>> Would it be valuable?
>>>>
>>>> One thing I understand is that there can be multiple discoverer's
>>>> for asteroid's & comets over a period of time (usually
>> within days
>>>> of each other).
>>>>
>>>> I have no knowledge on event modeling though. Do you have an
>>>> existing example or a model in mind?
>>>>
>>>> ~ Gordon
>>>>
>>>> <<< gordon at metaweb.com >>>
>>>>
>>>>
>>>>
>>>> On May 3, 2008, at 11:18 PM, Danny Hillis wrote:
>>>>
>>>>>
>>>>> On May 2, 2008, at 6:10 PM, Gordon Mackenzie wrote:
>>>>>>
>>>>>> So Celestial Object would have added specifically:
>>>>>>
>>>>>> Discoverer, Discovery Date, Discovery Method,
>> Discovery status,
>>>>>> Discovery Institution/organization
>>>>>>
>>>>>> And then remove/move the data for the related discovery
>>>> properties in
>>>>>> Planet/Asteroid/Comet  types to the equivalents in Celestial
>>> Body.
>>>>>> Planet type might have both jupiter and earth-based radius
>>>> and mass
>>>>>> properties.
>>>>>>
>>>>>
>>>>> Gordon,
>>>>> Have you thought about how the discovery of a planet get
>>>> modeled as an
>>>>> Event?
>>>>> -Danny
>>>> _______________________________________________
>>>> Data-modeling mailing list
>>>> Data-modeling at freebase.com
>>>> http://lists.freebase.com/mailman/listinfo/data-modeling
>>>>
>>>
>>> _______________________________________________
>>> Data-modeling mailing list
>>> Data-modeling at freebase.com
>>> http://lists.freebase.com/mailman/listinfo/data-modeling
>>
>> _______________________________________________
>> Data-modeling mailing list
>> Data-modeling at freebase.com
>> http://lists.freebase.com/mailman/listinfo/data-modeling
>>
>
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling