505 lines
14 KiB
YAML
505 lines
14 KiB
YAML
# ro.yaml
|
|
# -------
|
|
# Romanian language specification
|
|
|
|
components:
|
|
level:
|
|
# If no floor number is specified
|
|
null_probability: 0.6
|
|
alphanumeric_probability: 0.35
|
|
standalone_probability: 0.05
|
|
|
|
staircase:
|
|
null_probability: 0.95
|
|
alphanumeric_probability: 0.05
|
|
|
|
entrance:
|
|
null_probability: 0.999
|
|
alphanumeric_probability: 0.001
|
|
|
|
unit:
|
|
# If no unit number is specified
|
|
null_probability: 0.3
|
|
alphanumeric_probability: 0.65
|
|
standalone_probability: 0.05
|
|
|
|
numbers:
|
|
default: &numar
|
|
canonical: număr
|
|
abbreviated: nr
|
|
sample: true
|
|
canonical_probability: 0.1
|
|
abbreviated_probability: 0.7
|
|
sample_probability: 0.2
|
|
sample_exclude:
|
|
- "#"
|
|
numeric:
|
|
direction: left
|
|
numeric_affix:
|
|
affix: "#" # e.g. #3, #2F, etc.
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative:
|
|
direction: left # affix goes on the number's left
|
|
|
|
# Probabilities for numbers
|
|
numeric_probability: 0.9
|
|
numeric_affix_probability: 0.1
|
|
|
|
and:
|
|
default: &si
|
|
canonical: și
|
|
abbreviated: "&"
|
|
sample: true
|
|
canonical_probability: 0.5
|
|
abbreviated_probability: 0.4
|
|
sample_probability: 0.1
|
|
|
|
cross_streets:
|
|
and: *si
|
|
corner_of: &colt
|
|
canonical: colț
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
at_the_corner_of: &la_coltul_de_pe
|
|
canonical: la colțul de pe
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
intersection:
|
|
default: *si
|
|
probability: 0.7
|
|
alternatives:
|
|
- alternative: *colt
|
|
probability: 0.2
|
|
- alternative: *la_coltul_de_pe
|
|
probability: 0.1
|
|
|
|
between:
|
|
canonical: între
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
parentheses_probabililty: 0.5
|
|
|
|
|
|
house_numbers:
|
|
# fara numar (FN) addresses
|
|
no_number:
|
|
default:
|
|
canonical: fără număr
|
|
abbreviated: fn
|
|
sample: true
|
|
canonical_probability: 0.1
|
|
abbreviated_probability: 0.7
|
|
sample_probability: 0.2
|
|
alphanumeric:
|
|
default: *numar
|
|
|
|
alphanumeric_phrase_probability: 0.7
|
|
no_number_probability: 0.1 # With this probability, use fara numar if no house_number is specified
|
|
|
|
|
|
|
|
levels:
|
|
floor: &etaj
|
|
canonical: etaj
|
|
abbreviated: et
|
|
sample: true
|
|
canonical_probability: 0.5
|
|
abbreviated_probability: 0.4
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true # Occasionally add variation of "number", e.g. et. nr 2
|
|
add_number_phrase_probability: 0.05
|
|
digits:
|
|
ascii_probability: 0.8
|
|
roman_numeral_probability: 0.2
|
|
# Ground floor
|
|
parter: &parter
|
|
canonical: parter
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
aliases:
|
|
"0":
|
|
default: *parter
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *etaj
|
|
probability: 0.1
|
|
|
|
numbering_starts_at: 0
|
|
|
|
alphanumeric:
|
|
default: *etaj
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.05
|
|
numeric_probability: 0.99
|
|
alpha_probability: 0.01
|
|
|
|
blocks:
|
|
alphanumeric:
|
|
default:
|
|
canonical: bloc
|
|
abbreviated: bl
|
|
sample: true
|
|
canonical_probability: 0.6
|
|
abbreviated_probability: 0.2
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
|
|
|
|
categories:
|
|
near:
|
|
default:
|
|
canonical: in apropiere de
|
|
|
|
nearby:
|
|
default:
|
|
canonical: în apropiere
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative:
|
|
canonical: in apropiere
|
|
probability: 0.2
|
|
- alternative:
|
|
canonical: aproape de aici
|
|
probability: 0.1
|
|
- alternative:
|
|
canonical: aici
|
|
probability: 0.1
|
|
- alternative:
|
|
canonical: în jurul aici
|
|
probability: 0.05
|
|
- alternative:
|
|
canonical: in jurul aici
|
|
probability: 0.05
|
|
near_me:
|
|
default:
|
|
canonical: lângă mine
|
|
probability: 0.7
|
|
alternatives:
|
|
- alternative:
|
|
canonical: langa mine
|
|
probability: 0.3
|
|
in:
|
|
default:
|
|
canonical: din
|
|
# Probabilities of each phrase
|
|
near_probability: 0.35
|
|
nearby_probability: 0.2
|
|
near_me_probability: 0.1
|
|
in_probability: 0.35
|
|
|
|
directions:
|
|
right: &dreapta
|
|
canonical: dreapta
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: d
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.4
|
|
numeric_affix_probability: 0.6
|
|
left: &stanga
|
|
canonical: stânga
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: s
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.4
|
|
numeric_affix_probability: 0.6
|
|
|
|
alternatives:
|
|
- alternative: *dreapta
|
|
probability: 0.5
|
|
- alternative: *stanga
|
|
probability: 0.5
|
|
|
|
|
|
cardinal_directions:
|
|
east: &est
|
|
canonical: est
|
|
abbreviated: e
|
|
canonical_probability: 0.4
|
|
abbreviated_probability: 0.6
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: e
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
west: &vest
|
|
canonical: vest
|
|
abbreviated: v
|
|
canonical_probability: 0.4
|
|
abbreviated_probability: 0.6
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: v
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
north: &nord
|
|
canonical: nord
|
|
abbreviated: n
|
|
canonical_probability: 0.4
|
|
abbreviated_probability: 0.6
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: n
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
south: &sud
|
|
canonical: sud
|
|
abbreviated: s
|
|
canonical_probability: 0.4
|
|
abbreviated_probability: 0.6
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: s
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
alternatives:
|
|
- alternative: *nord
|
|
probability: 0.25
|
|
- alternative: *est
|
|
probability: 0.25
|
|
- alternative: *sud
|
|
probability: 0.25
|
|
- alternative: *vest
|
|
probability: 0.25
|
|
|
|
entrances:
|
|
entrada: &intrare
|
|
canonical: intrare
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
|
|
# Intrare 1, Intare A, etc.
|
|
alphanumeric:
|
|
default: *intrare
|
|
numeric_probability: 0.1 # e.g. Intrare 1
|
|
alpha_probability: 0.85 # e.g. Intrare 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
|
|
|
|
directional:
|
|
modifier:
|
|
alternatives:
|
|
- alternative: *nord
|
|
- alternative: *sud
|
|
- alternative: *est
|
|
- alternative: *vest
|
|
- alternative: *dreapta
|
|
- alternative: *stanga
|
|
|
|
staircases:
|
|
scara: &scara
|
|
canonical: scara
|
|
abbreviated: sc
|
|
sample: true
|
|
canonical_probability: 0.3
|
|
abbreviated_probability: 0.4
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
|
|
alphanumeric:
|
|
# For alphanumerics, Scara A, Scara 1, etc.
|
|
default: *scara
|
|
numeric_probability: 0.35 # e.g. Scara 1
|
|
alpha_probability: 0.6 # e.g. Scara 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
|
|
|
|
directional:
|
|
direction: right # e.g. Scara Nord
|
|
direction_probability: 0.8
|
|
modifier:
|
|
alternatives:
|
|
- alternative: *nord
|
|
- alternative: *sud
|
|
- alternative: *est
|
|
- alternative: *vest
|
|
- alternative: *dreapta
|
|
- alternative: *stanga
|
|
|
|
po_boxes:
|
|
casuta_postala: &casuta_postala
|
|
canonical: căsuță poștală
|
|
abbreviated: cp
|
|
sample: true
|
|
canonical_probability: 0.4
|
|
abbreviated_probability: 0.3
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.4 # Apdo No 1234
|
|
numeric_probability: 1.0
|
|
alphanumeric:
|
|
sample: false
|
|
default: *casuta_postala
|
|
numeric_probability: 0.9 # Apdo 123
|
|
alpha_probability: 0.05 # Apdo A
|
|
numeric_plus_alpha_probability: 0.04 # Apdo 123G
|
|
alpha_plus_numeric_probability: 0.01 # Apdo 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
|
|
|
|
units:
|
|
apartament: &apartament
|
|
canonical: apartament
|
|
abbreviated: ap
|
|
sample: true
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.6
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
sala: &sala
|
|
canonical: sală
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
birou: &birou
|
|
canonical: birou
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
lotul: &lotul
|
|
canonical: lotul
|
|
sample: true
|
|
canonical_probability: 0.7
|
|
sample_probability: 0.3
|
|
numeric:
|
|
direction: left
|
|
alphanumeric: &unit_alphanumeric
|
|
default: *apartament
|
|
probability: 0.9
|
|
sample: true
|
|
alternatives:
|
|
- alternative: *sala
|
|
probability: 0.1
|
|
|
|
# Separate random probability for adding directions like 2o Izq, 2 Dcha, etc.
|
|
add_direction: true
|
|
add_direction_probability: 0.1
|
|
add_direction_numeric: true # Only for numbers
|
|
|
|
numeric_probability: 0.9 # e.g. ap 1
|
|
numeric_plus_alpha_probability: 0.01 # e.g. ap 1A
|
|
alpha_plus_numeric_probability: 0.01 # e.g. ap A1
|
|
alpha_probability: 0.08 # e.g. ap A
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
|
|
zones:
|
|
residential: *unit_alphanumeric
|
|
commercial:
|
|
default: *birou
|
|
numeric_probability: 0.9 # e.g. Birou 1
|
|
numeric_plus_alpha_probability: 0.01 # e.g. Birou 1A
|
|
alpha_plus_numeric_probability: 0.01 # e.g. Birou A1
|
|
alpha_probability: 0.08 # e.g. Birou A
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
industrial:
|
|
default: *lotul
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative: *birou
|
|
probability: 0.3
|
|
- alternative: *sala
|
|
probability: 0.2
|
|
|
|
numeric_probability: 0.9 # e.g. Lotul 1
|
|
numeric_plus_alpha_probability: 0.01 # e.g. Lotul 1A
|
|
alpha_plus_numeric_probability: 0.01 # e.g. Lotul A1
|
|
alpha_probability: 0.08 # e.g. Lotul A
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
university:
|
|
default: *sala
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *birou
|
|
probability: 0.1
|
|
numeric_probability: 0.9 # e.g. Sala 1
|
|
numeric_plus_alpha_probability: 0.01 # e.g. Sala 1A
|
|
alpha_plus_numeric_probability: 0.01 # e.g. Sala A1
|
|
alpha_probability: 0.08 # e.g. Sala A
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|