552 lines
14 KiB
YAML
552 lines
14 KiB
YAML
# nl.yaml
|
|
# -------
|
|
# Note: base config covers Dutch as spoken in the Netherlands
|
|
# Belgium overrides go in country configs
|
|
|
|
components:
|
|
level:
|
|
null_probability: 0.85
|
|
alphanumeric_probability: 0.1
|
|
standalone_probability: 0.05
|
|
|
|
staircase:
|
|
null_probability: 0.99
|
|
alphanumeric_probability: 0.01
|
|
|
|
entrance:
|
|
null_probability: 0.999
|
|
alphanumeric_probability: 0.001
|
|
|
|
unit:
|
|
null_probability: 0.8
|
|
alphanumeric_probability: 0.2
|
|
|
|
combinations:
|
|
-
|
|
components:
|
|
- house_number
|
|
- unit
|
|
label: house_number
|
|
separators:
|
|
- separator: /
|
|
probability: 0.9
|
|
- separator: "-"
|
|
probability: 0.1
|
|
probability: 0.005
|
|
|
|
and:
|
|
default: &en
|
|
canonical: en
|
|
abbreviated: "&"
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.75
|
|
sample: true
|
|
sample_probability: 0.05
|
|
|
|
numbers:
|
|
default: &nummer
|
|
canonical: nummer
|
|
abbreviated: nr
|
|
sample: true
|
|
# Probabilities
|
|
canonical_probability: 0.3
|
|
abbreviated_probability: 0.5
|
|
sample_probability: 0.2
|
|
sample_exclude:
|
|
- "#"
|
|
numeric:
|
|
direction: left
|
|
numeric_affix:
|
|
affix: "#"
|
|
direction: left
|
|
|
|
numeric_probability: 0.4
|
|
numeric_affix_probability: 0.6
|
|
|
|
house_number:
|
|
alphanumeric:
|
|
default: *nummer
|
|
alphanumeric_phrase_probability: 0.01
|
|
|
|
levels:
|
|
verdieping: &verdieping
|
|
canonical: verdieping
|
|
sample: true
|
|
canonical_probability: 0.9
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
digits:
|
|
ascii_probability: 0.8
|
|
spellout_probability: 0.2
|
|
numeric_probability: 0.7
|
|
ordinal_probability: 0.3
|
|
etage: &etage
|
|
canonical: etage
|
|
abbreviated: et
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
abbreviated_probability: 0.1
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
digits:
|
|
ascii_probability: 0.8
|
|
spellout_probability: 0.2
|
|
numeric_probability: 0.7
|
|
ordinal_probability: 0.3
|
|
begane_grond: &begane_grond
|
|
canonical: begane grond
|
|
abbreviated: bg
|
|
sample: true
|
|
canonical_probability: 0.5
|
|
abbreviated_probability: 0.2
|
|
sample_probability: 0.3
|
|
benedenverdieping: &benedenverdieping
|
|
canonical: benedenverdieping
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
parterre: &parterre
|
|
canonical: parterre
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
gelijkvloers: &gelijkvloers
|
|
canonical: gelijkvloers
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
het_gelijkvloers: &het_gelijkvloers
|
|
canonical: het gelijkvloers
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
aliases:
|
|
"0":
|
|
default: *begane_grond
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: *benedenverdieping
|
|
probability: 0.35
|
|
- alternative: *parterre
|
|
probability: 0.04
|
|
- alternative: *het_gelijkvloers
|
|
probability: 0.005
|
|
- alternative: *gelijkvloers
|
|
probability: 0.005
|
|
alphanumeric:
|
|
default: *verdieping
|
|
probability: 0.99
|
|
alternatives:
|
|
- alternative: *etage
|
|
probability: 0.01
|
|
numeric_probability: 0.99 # With this probability, pick an integer
|
|
alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
|
|
numeric_plus_alpha_probability: 0.0001 # e.g. 2A
|
|
alpha_plus_numeric_probability: 0.0001 # e.g. A2
|
|
|
|
categories:
|
|
near:
|
|
default:
|
|
canonical: in de buurt van
|
|
probability: 0.8
|
|
alternatives:
|
|
- alternative:
|
|
canonical: bij
|
|
probability: 0.1
|
|
- alternative:
|
|
canonical: nabij
|
|
probability: 0.1
|
|
nearby:
|
|
default:
|
|
canonical: in de buurt
|
|
near_me:
|
|
default:
|
|
canonical: in de buurt van me
|
|
|
|
in:
|
|
default:
|
|
canonical: in
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative:
|
|
canonical: te
|
|
probability: 0.4
|
|
# Probabilities of each phrase
|
|
near_probability: 0.35
|
|
nearby_probability: 0.2
|
|
near_me_probability: 0.1
|
|
in_probability: 0.35
|
|
|
|
|
|
|
|
cross_streets:
|
|
and: *en
|
|
corner_of: &hoek_van
|
|
canonical: hoek van
|
|
at_the_corner_of: &op_de_hoek_van
|
|
canonical: op de hoek van
|
|
intersection:
|
|
default: *en
|
|
probability: 0.7
|
|
alternatives:
|
|
- alternative: *hoek_van
|
|
probability: 0.15
|
|
- alternative: *op_de_hoek_van
|
|
probability: 0.15
|
|
|
|
between:
|
|
canonical: tussen
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
parentheses_probability: 0.5
|
|
|
|
|
|
entrances:
|
|
ingang: &ingang
|
|
canonical: ingang
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
# Eingang 1, Eingang A, etc.
|
|
alphanumeric: &entrance_alphanumeric
|
|
default: *ingang
|
|
numeric_probability: 0.1 # e.g. Eingang 1
|
|
alpha_probability: 0.85 # e.g. Eingang A
|
|
numeric_plus_alpha_probability: 0.025 # e.g. 1A
|
|
alpha_plus_numeric_probability: 0.025 # e.g. A1
|
|
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
po_boxes:
|
|
postbus: &postbus
|
|
canonical: postbus
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.2
|
|
antwoordnummer: &antwoordnummer
|
|
canonical: antwoordnummer
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.2
|
|
alphanumeric:
|
|
sample: false
|
|
default: *postbus
|
|
probability: 0.8
|
|
alternatives:
|
|
- alternative: *antwoordnummer
|
|
probability: 0.2
|
|
numeric_probability: 0.9 # 123
|
|
alpha_probability: 0.05 # A
|
|
numeric_plus_alpha_probability: 0.04 # 123G
|
|
alpha_plus_numeric_probability: 0.01 # A123
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
digits:
|
|
- length: 1
|
|
probability: 0.05
|
|
- length: 2
|
|
probability: 0.1
|
|
- length: 3
|
|
probability: 0.2
|
|
- length: 4
|
|
probability: 0.5
|
|
- length: 5
|
|
probability: 0.1
|
|
- length: 6
|
|
probability: 0.05
|
|
|
|
directions:
|
|
right: &rechts
|
|
canonical: rechts
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: r
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.8
|
|
numeric_affix_probability: 0.2
|
|
left: &links
|
|
canonical: links
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: l
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.4
|
|
numeric_affix_probability: 0.6
|
|
alternatives:
|
|
- alternative: *rechts
|
|
probability: 0.5
|
|
- alternative: *links
|
|
probability: 0.5
|
|
|
|
|
|
cardinal_directions:
|
|
east: &oost
|
|
canonical: oost
|
|
abbreviated: o
|
|
canonical_probability: 0.95
|
|
abbreviated_probability: 0.05
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: o
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
oosten: &oosten
|
|
<<: *oost
|
|
canonical: oosten
|
|
|
|
oostelijke: &oostelijke
|
|
canonical: oostelijke
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
west: &west
|
|
canonical: west
|
|
abbreviated: w
|
|
canonical_probability: 0.95
|
|
abbreviated_probability: 0.05
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: w
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
westen: &westen
|
|
<<: *west
|
|
canonical: westen
|
|
|
|
westelijke: &westelijke
|
|
canonical: westelijke
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
north: &noord
|
|
canonical: noord
|
|
abbreviated: n
|
|
canonical_probability: 0.95
|
|
abbreviated_probability: 0.05
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: n
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
noorden: &noorden
|
|
<<: *noord
|
|
canonical: noorden
|
|
|
|
noordelijke: &noordelijke
|
|
canonical: noordelijke
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
south: &zuid
|
|
canonical: zuid
|
|
abbreviated: z
|
|
sample: true
|
|
canonical_probability: 0.75
|
|
abbreviated_probability: 0.1
|
|
sample_probability: 0.15
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: z
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
zuiden: &zuiden
|
|
<<: *zuid
|
|
canonical: zuiden
|
|
|
|
zuidelijke: &zuidelijke
|
|
canonical: zuidelijke
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
alternatives:
|
|
- alternative: *noord
|
|
probability: 0.25
|
|
- alternative: *oost
|
|
probability: 0.25
|
|
- alternative: *zuid
|
|
probability: 0.25
|
|
- alternative: *west
|
|
probability: 0.25
|
|
|
|
|
|
staircases:
|
|
stiege: &stiege
|
|
canonical: stiege
|
|
abbreviated: stg
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
abbreviated_probability: 0.2
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: left
|
|
trap: &trap
|
|
canonical: trap
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
alphanumeric: &staircase_alphanumeric
|
|
default: *trap
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: *stiege
|
|
probability: 0.4
|
|
numeric_probability: 0.75
|
|
alpha_probability: 0.2
|
|
numeric_plus_alpha_probability: 0.025
|
|
alpha_plus_numeric_probability: 0.025
|
|
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
units:
|
|
appartement: &appartement
|
|
canonical: appartement
|
|
abbreviated: apt
|
|
sample: true
|
|
canonical_probability: 0.3
|
|
abbreviated_probability: 0.5
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
kamer: &kamer
|
|
canonical: kamer
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
alphanumeric: &unit_alphanumeric
|
|
default: *appartement
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: *kamer
|
|
probability: 0.4
|
|
numeric_probability: 0.9 # e.g. Apt 1
|
|
numeric_plus_alpha_probability: 0.03 # e.g. 1A
|
|
alpha_plus_numeric_probability: 0.03 # e.g. A1
|
|
alpha_probability: 0.04 # e.g. Apt A
|
|
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
# Separate random probability for adding directions like 2R, 2L, etc.
|
|
add_direction: true
|
|
add_direction_probability: 0.1
|
|
|
|
# Add directions for plain numbers
|
|
add_direction_numeric: true
|
|
# Add direction only e.g. Apt Rechts
|
|
add_direction_standalone: true
|
|
|
|
# If there are 10 floors, create unit numbers like #301 or #1032
|
|
use_floor_probability: 0.1
|
|
|
|
|
|
countries:
|
|
be:
|
|
components:
|
|
unit:
|
|
null_probability: 0.65
|
|
alphanumeric_probability: 0.35
|
|
|
|
levels:
|
|
verdieping: &verdieping_flemish
|
|
canonical: verdieping
|
|
abbreviated: verdiep
|
|
sample: true
|
|
canonical_probability: 0.3
|
|
abbreviated_probability: 0.5
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
digits:
|
|
ascii_probability: 0.8
|
|
spellout_probability: 0.2
|
|
numeric_probability: 0.7
|
|
ordinal_probability: 0.3
|
|
|
|
aliases:
|
|
"0":
|
|
default: *het_gelijkvloers
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative: *gelijkvloers
|
|
probability: 0.5
|
|
alphanumeric:
|
|
default: *verdieping_flemish
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *etage
|
|
probability: 0.1
|
|
|
|
units:
|
|
bus: &bus
|
|
canonical: bus
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
alphanumeric:
|
|
default: *appartement
|
|
probability: 0.1
|
|
alternatives:
|
|
- alternative: *bus
|
|
probability: 0.7
|
|
- alternative: *kamer
|
|
probability: 0.2
|