Repository files navigation

Where things are (and what they mean) in Who's On First

For any hierarchy of place types we have identified three “roles” that any one of those place types can fall into. That doesn't mean there can't be others (roles or place types) only that these are the ones we've identified as a good place to start.

These are, well, common acrossany hierarchy for any place in Who's On First.

This part is important: It means that at some point every single record shares at least one or more common ancestors (for example a country or a continent or occasionally just the Earth). That doesn't preclude very specific additions to the hierarchy for a given location only that those additions need to fit within a common hierarchy shared across all locations.

These are meant to be part of a common hierarchy but may not be present because they aren't relevant or because we don't have the data. Counties are a good example of this.

These are the parts of a hierarchy specific, typically, to a country or region. For example, the many nested "departments" in France or Germany. The only rule is that an optional (O) place type has to fit somewhere inside the common (C) hierarchy.

So the minimum list of place types for a hierarchy applied globally looks like this:

- continent (C)  - country (C)    - region (C)       - county (CO)          - locality (C)            - neighbourhood (C)

A more nuanced version might look like this:

- continent (C)  - empire (CO)    - country (C)     - macroregion (O)      - region (C)       - macrocounty (O)        - county (CO)         - metro area (CO)          - locality (C)           - macrohood (O)            - neighbourhood (C)             - microhood (O)              - campus  (CO)               - building (CO)                - address (CO)                 - venue (C)

The blue boxes arecommon placetypes, the green onescommon_optional and the yellow ones areoptional.

This image was generated using thehandy wof-graph-placetypes script in the py-mapzen-whosonfirst-placetypes library

All place types have a unique 64-bit numeric ID. It should always be possible to query or filter for places by that numeric ID so that the burden of remembering whether it isneighbourhood orneighborhood orquartier or whatever is reserved for friendly banter over drinks.

In alphabetical order the current list of placetypes in Who's On First is:

102312329

This is an open question. The question being: How/what do we do with OpenAddresses? We aren't sure of anything, yet, except maybe to put this here as a placeholder for probably between buildings and venues

Addresses are consideredcommon_optional.

1159162575

Arcades are consideredoptional.

421205763

A local government unit, below thelocality placetype. ThinkBrooklyn.

Boroughs are consideredcommon_optional.

102312327

See also:Imagining the Built Works Registry

Buildings are consideredcommon_optional.

102312331

Things like universities or office complexes andairports. Something something something, Burning Man and other annual events.

Campuses are consideredcommon_optional.

1159162573

Concourses are consideredoptional.

1108746739

An area entitled to elect a representative.

Constituencies are consideredcommon_optional.

102312309

I think we're all in pretty broad agreement about continents, yeah?

Continents are consideredcommon.

102312313

This needs a better - that is more abstract - name. Like "region" instead of state, province, whatever... but for counties.

Counties are consideredcommon_optional.

102312307

Basically places that issue passports, notwithstanding the details (like empires which actually issue the passports...). In some areas country and dependency are used interchangeably.

Continents are consideredcommon.

1729783759

Custom placetypes are provided as a convenience for projects producing data that live outside of the "core" Who's On First (WOF) data repositories as a way to allow those records to have a well-known WOF placetype. This ensures that these projects can continue to use the WOF-related tooling, some of which enforces strict validation on things like placetypes.

Thecustom placetype has no predefined list of ancestors or any other computationally-derived metadata so properties likewof:hierarchy will need to be produced by the projects using thecustom placetype on an as-needed basis. Any specific detail about a record's placetype should be encoded in a project-specific namespace.

Custom placetypes are consideredoptional.

102312313

It's not a sub-region of a country but rather dependent on a parent country for defence, passport control, subsidies, etc. In some areas country and dependency are used interchangeably.

Dependencies are consideredcommon_optional.

102322043

Places that one or more parties claim as their own. As of this writingall disputed places are parented only by the country (and higher) IDs of the claimants. This isn't to say there aren't more granular hierarchies to be applied to these place only that we are starting with the simple stuff first.

Disputed areas are consideredcommon_optional.

102312335

Or "sovereignty" but really... empire. For example the Meta United States that contains both the US and Puerto Rico.

Empires are consideredcommon_optional.

1159268867

Enclosures are consideredoptional.

1159268869

Installations are consideredoptional.

1108906905

Where things meet.

Intersections are consideredoptional.

404221409

In many countries, the lowest level of government. They contain one or more localities (or "populated places") which themselves have no authority. Often but not exclusively found in Europe. In some areas locality and localadmin are used interchangeably.

Localadmins are consideredcommon_optional.

102312317

Towns and cities, independent of size or population. Things with neighbourhoods, basically. In some areas locality and localadmin are used interchangeably.

Localities are consideredcommon.

404221413

Bundles of counties! These exists mostly in Europe.

Macrocounties are consideredoptional.

102312323

Like "BoCoCa" which in WOE is a neighbourhood that parents another... neighbourhood.

Macrohoods are consideredoptional.

404221411

Bundles of regions! These exists mostly in Europe.

Macroregions are consideredoptional.

404528655

Places with fish and boats.

Marine areas are consideredcommon_optional.

1360666019

Economic geography areas that are usually an aggregation of counties or postalcodes based on TV, radio, and other media related audiences.

Market areas are consideredoptional.

102371933

Things like "The Bay Area" – this one is hard so we shouldn't spend too much time worrying about the details yet but instead treat as something we want to do eventually.

Metropolitain areas are consideredcommon_optional.

102312321

Because all place is disputed. And everyone has a name for a place that will offend someone else.

Microhoods are consideredoptional.

1796730315

Nations are things that might otherwise be considered "countries" (in the past) but because the semantics of grouping these with contemporary (administrative state) countries would be confusing a separate placetype was added.

One way to think about is that all countries are nations but, in Who's On First, not all nations are countries. I know, right?

Nations are consideredoptional.

102312319

Neighbourhoods can represent an area surrounding a "place", a community grouping, or even a physical area. One thing is certain - neighbourhoods are things no one will completely agree on. Ever.

Neighbourhoods are consideredcommon.

404528653

The really big marine areas. The ones that you sometimes feel cast adrift on...

Oceans are consideredcommon_optional.

102312341

The big round thing you're standing on.

The planet is consideredcommon_optional.

470996387

Used for mail sorting and stringing addresses together.

Postalcodes are consideredcommon_optional.

1897483759

Used for denoting areas for mail sorting. Often a prefix ofpostalcode, but not always.

Postalregions are consideredcommon_optional.

102312311

States, provinces, regions. We call themregions. These are places that would have a bone in a "states rights" argument.

Regions are consideredcommon.

136057795

An area where a single, standard time is used.

Timezones are consideredcommon_optional.

102312325

Things with walls, often but mostly things that people stand aroundtogether. Things with walls might be public (a bar) or private (your apartment) by default.

Venues are consideredcommon_optional.

1159162571

Wings are consideredoptional.

Per the ISO 3166 spec which states:

User-assigned code elements are codes at the disposal of users who need to add further names of countries, territories, or other geographical entities to their in-house application of ISO 3166-1, and the ISO 3166/MA will never use these codes in the updating process of the standard. The following codes can be user-assigned:[19]    Alpha-2: AA, QM to QZ, XA to XZ, and ZZ    Alpha-3: AAA to AAZ, QMA to QZZ, XAA to XZZ, and ZZA to ZZZ    Numeric: 900 to 999

We use the following ISO country codes:

We just followedGeonames' lead and have assignedXK to be the ISO country code for Kosovo.

ForNull Island.

We useXS to indicate Somaliland.

XX denotes a place disputed by two or more (ISO) countries.

XY denotes an ISO country that has yet to be determined (by us). You might typically see this is a record for a freshly created place that hasn't been fully vetted or editorialized yet.

XZ is the ISO country code equivalent ofwof:parent_id=-2 or:shrug: the world is a complicated place.

A Who's On First (wof:) hierarchy is a list of dictionaries, which each item is a dictionary containing a full hierarchy. Like this:

"wof:hierarchy": [{ "neighbourhood_id": 9997, "locality_id": 9997, "metro_id": 9998, "county_id": 9998, "region_id": 9998, "country_id": 9998, "continent_id": 9998 },{ "neighbourhood_id": 9997, "locality_id: 9997, "metro_id": 9999, "county_id": 9999, "region_id": 9999, "country_id": 9999, "continent_id": 9999 }]

See below for details and rationale.

The following is a verbatim written exercise to work through the issue of how a hierarchy should be represented or, more specifically, how the potentially multiple hierarchies that a given place might encompass should be represented.

Something something something as elements on the rootproperties dictionary. Like this:

{"wof:neighbourhood_id": 9999,"wof:locality_id": 9999,"wof:county_id": 9999,"wof:region_id": 9999,"wof:country_id": 9999,"wof:continent_id": 9999}

Something something something as a dictionary on the rootproperties dictionary. Like this:

"wof:hierarchy": {"neighbourhood_id": 9999,"locality_id": 9999,"county_id": 9999,"region_id": 9999,"country_id": 9999,"continent_id": 9999}

This has the advantage of keeping all the hierarchy information in one place however it is left as an exercise to the user to enforce the actual order of the hierarchy since there is no way to be certain thatprogramming langage X will ensure the ordering of the dictionary keys. We could define the hierarchy as a list of dictionaries which would allow to explicitly encode the parent for that node but those extra bytes in each record will add up fast when dealing with a global hierarchy. Like this:

"wof:hierarchy": [{ "neighbourhood_id": 9999, "parent": "locality_id" },{ "locality_id": 9999, "parent: "county_id" },{ "county_id": 9999, "parent": "region_id" },{ "region_id": 9999, "parent": "country_id" },{ "country_id": 9999, "parent": "continent_id" },{ "continent_id": 9999, "parent": "" }}

Meanwhile we also know that we want to support certain place types that will have multiple parents (because geography) like metropolitain areas or, if we choose to include them in the gazetteer proper, road networks.

At a minimum this means that some of the values for placetypes have to be lists which probably meansall of the values should be lists so that people don't have to think about context or test data types. For example:

"wof:hierarchy": {"neighbourhood_id": [ 9999 ],"locality_id": [ 9999 ],"metropolitain_area": [ 9999 ],"county_id": [ 9999, 9999 ],"region_id": [ 9999 ],"country_id": [ 9999 ],"continent_id": [ 9999 ],}

However, it is possible to imagine a place type with not only multiple parents but multiple ancestors. A timezone or, again, a road network. In which case you find yourself with a dictionary whose values are lists of dictionaries. At which point you risk spiralling off in to Semantic Web graph theory quicksand.

So maybe the thing to do is suffer mixed content (unique IDs and lists) where the rule is the immediate (outer) hierarchy stops the moment there are multiple parents. Like this:

"wof:hierarchy": {"neighbourhood_id": 9999,"locality_id": 9999,"metro_id": [{ "metro_id": 9998, "county_id": 9998, "region_id": 9998, "country_id": 9998, "continent_id": 9998 },{ "metro_id": 9999, "county_id": 9999, "region_id": 9999, "country_id": 9999, "continent_id": 9999 }]}

Or maybe the hierarchy is always just a list of dictionaries, each containing a full hierarchy. Like this:

"wof:hierarchy": [{ "neighbourhood_id": 9997, "locality_id": 9997, "metro_id": 9998, "county_id": 9998, "region_id": 9998, "country_id": 9998, "continent_id": 9998 },{ "neighbourhood_id": 9997, "locality_id: 9997, "metro_id": 9999, "county_id": 9999, "region_id": 9999, "country_id": 9999, "continent_id": 9999 }]

Reasons why this last suggestion is good:

• It is explicit
• It is easy to compare multiple hierarchies
• It doesn't require the user do a lot of mental arithmetic to construct the complete hierarchy or to support whatever "efficiencies" we dream up in the moment
• It is easier to change going forward (say before an "official" launch) than the alternatives

Reasons why this last suggestion is, or might be, bad:

• If we support metropolitain areas then many places (localities, neighbourhood, venues) may have multiple hierarchies where the only difference will (likely) be the county, leaving all the remaining ancestors in common
• File size, disk space and bandwidth - this is the corollary of the first point and akin to whitespace or coordinates with > 6 decimal points in GeoJSON files

In the end the "good" reasons outweighed the "bad" reasons.

• https://github.com/whosonfirst/whosonfirst-docs
• https://github.com/whosonfirst/whosonfirst-placetypes-constituencies
• https://github.com/whosonfirst/go-whosonfirst-placetypes