471 lines
12 KiB
YAML
471 lines
12 KiB
YAML
# et.yaml
|
|
# -------
|
|
# Estonian language specification.
|
|
|
|
components:
|
|
level:
|
|
null_probability: 0.97
|
|
alphanumeric_probability: 0.02
|
|
standalone_probability: 0.01
|
|
|
|
staircase:
|
|
null_probability: 0.99
|
|
alphanumeric_probability: 0.01
|
|
|
|
entrance:
|
|
null_probability: 0.999
|
|
alphanumeric_probability: 0.001
|
|
|
|
unit:
|
|
null_probability: 0.75
|
|
alphanumeric_probability: 0.25
|
|
|
|
combinations:
|
|
-
|
|
components:
|
|
- house_number
|
|
- unit
|
|
label: house_number
|
|
separators:
|
|
- separator: "-"
|
|
probability: 0.95
|
|
- separator: " - "
|
|
probability: 0.05
|
|
probability: 0.7
|
|
|
|
|
|
numbers:
|
|
default: &number
|
|
canonical: number
|
|
abbreviated: nbr
|
|
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_numbers:
|
|
alphanumeric:
|
|
default: *number
|
|
|
|
alphanumeric_phrase_probability: 0.0001
|
|
|
|
|
|
and:
|
|
default: &ja
|
|
canonical: ja
|
|
abbreviated: "&"
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.75
|
|
sample: true
|
|
sample_probability: 0.05
|
|
|
|
cross_streets:
|
|
and: *ja
|
|
corner_of: &nurgas
|
|
canonical: nurgas
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
at_the_corner_of: &nurgal
|
|
canonical: nurgal
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
intersection:
|
|
default: *ja
|
|
probability: 0.7
|
|
alternatives:
|
|
- alternative: *nurgas
|
|
probability: 0.15
|
|
- alternative: *nurgal
|
|
probability: 0.15
|
|
|
|
between:
|
|
canonical: vahel
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
parentheses_probability: 0.5
|
|
|
|
levels:
|
|
floor: &korrusel
|
|
canonical: korrusel
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
direction_probability: 0.9
|
|
ordinal:
|
|
direction: right
|
|
numeric_probability: 0.4
|
|
ordinal_probability: 0.6
|
|
parter: &parter
|
|
canonical: parter
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
kelder: &kelder
|
|
canonical: kelder
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
standalone_probability: 1.0
|
|
keldris: &keldris
|
|
canonical: keldris
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
# e.g. 1 keldris
|
|
numeric:
|
|
direction: right
|
|
direction_probability: 0.8
|
|
# e.g. k1
|
|
numeric_affix:
|
|
affix: k
|
|
direction: left
|
|
# e.g. 1. keldris
|
|
ordinal:
|
|
direction: right
|
|
number_abs_value: true
|
|
number_min_abs_value: 1
|
|
numeric_probability: 0.05
|
|
numeric_affix_probability: 0.9
|
|
ordinal_probability: 0.05
|
|
aliases:
|
|
"<-1":
|
|
default: *kelder
|
|
probability: 0.85
|
|
alternatives:
|
|
- alternative: *keldris
|
|
probability: 0.15
|
|
"-1":
|
|
default: *kelder
|
|
probability: 0.85
|
|
alternatives:
|
|
- alternative: *keldris
|
|
probability: 0.1
|
|
- alternative: *korrusel
|
|
probability: 0.05
|
|
"1":
|
|
default: *parter
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative: *korrusel
|
|
probability: 0.5
|
|
|
|
numbering_starts_at: 1
|
|
|
|
alphanumeric:
|
|
default: *korrusel
|
|
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: lähedal
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
nearby:
|
|
default:
|
|
canonical: lähedal
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
probability: 0.7
|
|
alternatives:
|
|
- alternative:
|
|
canonical: siin lähedal
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
probability: 0.2
|
|
- alternative:
|
|
canonical: siinkandis
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
probability: 0.1
|
|
|
|
near_me:
|
|
default:
|
|
canonical: lähedal mulle
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
|
|
# Probabilities of each phrase
|
|
near_probability: 0.7
|
|
nearby_probability: 0.2
|
|
near_me_probability: 0.1
|
|
|
|
directions:
|
|
right: &paremal
|
|
canonical: paremal
|
|
sample: true
|
|
canonical_probability: 0.9
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: p
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.8
|
|
numeric_affix_probability: 0.2
|
|
paramale: &paremale
|
|
canonical: paremale
|
|
sample: true
|
|
canonical_probability: 0.9
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: p
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.8
|
|
numeric_affix_probability: 0.2
|
|
left: &vasakul
|
|
canonical: vasakul
|
|
sample: true
|
|
canonical_probability: 0.9
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: v
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.8
|
|
numeric_affix_probability: 0.2
|
|
vasakule: &vasakule
|
|
canonical: vasakule
|
|
sample: true
|
|
canonical_probability: 0.9
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: v
|
|
direction: right
|
|
whitespace_probability: 0.1
|
|
numeric_probability: 0.8
|
|
numeric_affix_probability: 0.2
|
|
alternatives:
|
|
- alternative: *paremal
|
|
probability: 0.25
|
|
- alternative: *paremale
|
|
probability: 0.25
|
|
- alternative: *vasakul
|
|
probability: 0.25
|
|
- alternative: *vasakule
|
|
probability: 0.25
|
|
|
|
cardinal_directions:
|
|
east: &ida
|
|
canonical: ida
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
west: &laas
|
|
canonical: lääs
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
north: &pohi
|
|
canonical: põhi
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
|
|
south: &louna
|
|
canonical: lõuna
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
|
|
alternatives:
|
|
- alternative: *pohi
|
|
probability: 0.25
|
|
- alternative: *ida
|
|
probability: 0.25
|
|
- alternative: *louna
|
|
probability: 0.25
|
|
- alternative: *laas
|
|
probability: 0.25
|
|
|
|
|
|
entrances:
|
|
sissepaas: &sissepaas
|
|
canonical: sissepääs
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
# Eingang 1, Eingang A, etc.
|
|
alphanumeric: &entrance_alphanumeric
|
|
default: *sissepaas
|
|
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
|
|
|
|
staircases:
|
|
trepikoda: &trepikoda
|
|
canonical: trepikoda
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
alphanumeric: &staircase_alphanumeric
|
|
default: *trepikoda
|
|
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
|
|
|
|
directional:
|
|
direction: left
|
|
direction_probability: 0.85
|
|
modifier:
|
|
alternatives:
|
|
- alternative: *pohi
|
|
- alternative: *louna
|
|
- alternative: *ida
|
|
- alternative: *laas
|
|
|
|
po_boxes:
|
|
postboks: &abonementpostkast
|
|
canonical: abonementpostkast
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.2 # abonementpostkast #1234
|
|
kast: &kast
|
|
canonical: kast
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.2 # Kast #1234
|
|
alphanumeric:
|
|
sample: false
|
|
default: *abonementpostkast
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *kast
|
|
probability: 0.1
|
|
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
|
|
|
|
units:
|
|
korter: &korter
|
|
canonical: korter
|
|
abbreviated: k
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
abbreviated_probability: 0.1
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: left
|
|
null_phrase_probability: 0.3
|
|
# Lejlighed nummer 4
|
|
add_number_phrase: true
|
|
add_number_phrase_probability: 0.05
|
|
ruumi: &ruumi
|
|
canonical: ruumi
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
alphanumeric: &unit_alphanumeric
|
|
default: *korter
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *ruumi
|
|
probability: 0.1
|
|
numeric_probability: 1.0 # e.g. korter 1
|
|
|
|
# Separate random probability for adding directions like 2P, 2V, etc.
|
|
add_direction: true
|
|
add_direction_probability: 0.005
|
|
|
|
# Add directions for plain numbers
|
|
add_direction_numeric: true
|
|
# Add direction only e.g. Korter vasakule
|
|
add_direction_standalone: true
|
|
|
|
# If there are 10 floors, create unit numbers like #301 or #1032
|
|
use_floor_probability: 0.05
|