tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[places] Place config with dropout probabilities for admin boundaries by country, exceptions by OSM relation id

Browse Source
This commit is contained in:
Al
2016-05-19 03:18:06 -04:00
parent 4c109317b1
commit dfaad89c9d
3 changed files with 292 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

201
resources/places/countries/global.yaml Normal file
View File

@@ -0,0 +1,201 @@
# This config specifies isolated and conditional probabilities
# of including various admin components (applies to both addresses
# and standalone place queries)
global:
# Probability of including individual components
components:
suburb:
probability: 0.2
city_district:
probability: 0.2
city:
probability: 0.85
island:
# Islands are usually not represented in addresses
probability: 0.0
state_district:
probability: 0.05
state:
probability: 0.1
country:
probability: 0.6
countries:
# Australia
au:
components:
state:
probability: 0.6
# Canada
ca:
components:
state:
probability: 0.7
# France
fr:
components:
city_district:
probability: 0.4
# United Kingdom
gb:
components:
suburb:
probability: 0.5
# Indonesia
id:
components:
island:
probability: 0.4
# Hong Kong
hk:
components:
country:
probability: 0.85
state:
probability: 0.2
island:
probability: 0.4
containing:
- id: 2278450 # Hong Kong Island (already a "state")
type: relation
probability: 0.0
# Japan
jp:
components:
island:
probability: 0.4
suburb:
probability: 0.6
city_district:
probability: 0.8
state:
probability: 0.8
country:
probability: 0.4
hu:
components:
country:
probability: 0.1
# Saint Kitts and Nevis
kn:
components:
island:
probability: 0.8
# Malaysia
my:
components:
island:
probability: 0.3
# Pitcairn Islands
pn:
components:
island:
probability: 0.8
# Seychelles
sc:
components:
island:
probability: 0.8
# United States
us:
# Definitions
kings_county: &kings_county
id: 369518 # Kings County (Brooklyn, NY)
type: relation
queens_county: &queens_county
id: 369519 # Queens County (Queens, NY)
type: relation
bronx_county: &bronx_county
id: 2552450 # Bronx County (Bronx, NY)
type: relation
richmond_county: &richmond_county
id: 962876 # Richmond County (Staten Island, NY)
type: relation
hawaii: &hawaii
id: 166563 # State of Hawaii
type: relation
components:
suburb:
probability: 0.4
city_district:
probability: 0.2
containing:
- <<: *kings_county
probability: 0.85
- <<: *queens_county
probability: 0.85
- <<: *bronx_county
probability: 0.85
- <<: *richmond_county
probability: 0.85
city:
containing:
- <<: *kings_county
probability: 0.1
- <<: *queens_county
probability: 0.1
- <<: *bronx_county
probability: 0.1
- <<: *richmond_county
probability: 0.1
island:
order:
direction: before
component: state_district
containing:
# Island is more common in Hawaiian addresses
- <<: *hawaii
probability: 0.8
state:
probability: 0.7
# Higher probability of Brooklyn/city_district NY/state
containing:
- <<: *kings_county
probability: 0.8
- <<: *queens_county
probability: 0.8
- <<: *bronx_county
probability: 0.8
- <<: *richmond_county
probability: 0.8
state_district:
probability: 0.1
country:
probability: 0.1
# Tuvalu
tv:
components:
island:
probability: 0.8
# US Virgin Islands
vi:
components:
island:
probability: 0.8
# British Virgin Islands
vg:
components:
island:
probability: 0.8