[Data-modeling] Geosearch service (was Re: Modeling uncertainty)
Jamie Taylor
jamie at metaweb.com
Thu Feb 26 19:20:41 UTC 2009
Tom et al. -
I'm coming to this thread somewhat late, but I just wanted to note
that the geosearch API is (rather) experimental. We are still working
out the specific capabilities of the service and may rework the API to
reflect that. That said, combining geo/semantics seems very powerful,
so we are interested in seeing how people use the current incarnation
of the service and what usecases people are trying to address.
J
On Feb 26, 2009, at 10:47 AM, Faye Harris wrote:
> Hi Tom,
>
> My post was meant to point out that when it comes to modeling
> geographical data, there is established convention regarding
> concepts of
> "near" or proximity. A "place" can be expressed as a point or a shape,
> which the query language naturally supports. Modeling this data using
> the same convention would result in better ability for retrieval and
> analysis down the road.
>
> -- Faye
>
>
> Tom Morris wrote:
>> Thanks Faye. I was aware of that service and had explicitly
>> searched,
>> unsuccessfully, for information on it, so I appreciate the pointer.
>> You might want to add it to
>> https://www.freebase.com/view/freebase/metaweb_api_service and
>> http://www.freebase.com/view/guid/9202a8c04000641f80000000092a10a0
>>
>> While I'll definitely check out the new service, my original comment
>> was about the modeling side of things, not the query side of things.
>> For example, the National Park Service's National Register of
>> Historic
>> Places has a boolean flag on their locations called "Vicinity." If
>> this flag is set True in their database, there's no way to record
>> that
>> location in Metaweb. It's also not uncommon in the historical record
>> to have locations recorded imprecisely saying, for example, that
>> someone was from "near Oxford" when they lived in an outlying village
>> or farm. Recording the location as "Oxford" is inaccurate, but the
>> alternative is to just drop the location altogether and lose
>> information.
>>
>> Tom
>>
>> On Wed, Feb 25, 2009 at 8:17 PM, Faye Harris <faye at metaweb.com>
>> wrote:
>>
>>> Hi Tom,
>>>
>>> This just caught my attention.
>>>
>>> For location-related queries, are you aware of Freebase's
>>> /api/service/geosearch? Given a search query and some geographical
>>> requirement, it returns you results that match your search phrase
>>> and
>>> fit your geographical requirement. Specifically, you can use the
>>> parameter "within" to pass in "a floating pointing number denoting
>>> the
>>> radius in kilometers searched from the anchor location".
>>>
>>> For example, the following query returns restaurants within 20km of
>>> Boston (city center):
>>> http://freebase.com/api/service/geosearch?location=/en/boston_massachusetts&type=/dining/restaurant&within=20&indent=1
>>>
>>> Does that meet your needs? Documentation for /api/service/
>>> geosearch is
>>> available at:
>>> http://freebase.com/api/service/geosearch?help
>>>
>>> -- Faye
>>>
>>>
>>> Tom Morris wrote:
>>>
>>>> Location - If I'm told that something is "near" or "in the vicinity
>>>> of" a location, currently my choices are to either not record the
>>>> fact
>>>> or to guess at a way to reduce the precision in a way that's still
>>>> accurate. I could say that something which is "near Boston" is "in
>>>> Massachusetts," but a) that might not be true and b) that's not the
>>>> information that I have.
>>>>
>>> _______________________________________________
>>> 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
More information about the Data-modeling
mailing list