595 lines
16 KiB
YAML
595 lines
16 KiB
YAML
# nb.yaml
|
||
# -------
|
||
# Norwegian language specification.
|
||
|
||
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: 0.75
|
||
alphanumeric_probability: 0.25
|
||
|
||
combinations:
|
||
# e.g. 2/34, more common way to specify a unit number in German
|
||
# if unit exists in the first place
|
||
level_unit:
|
||
components:
|
||
- level
|
||
- unit
|
||
label: unit
|
||
zero_pad_digits: 2
|
||
separators:
|
||
- separator: ""
|
||
probability: 1.0
|
||
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: 0.0001
|
||
|
||
|
||
and:
|
||
default: &og
|
||
canonical: og
|
||
abbreviated: "&"
|
||
canonical_probability: 0.2
|
||
abbreviated_probability: 0.75
|
||
sample: true
|
||
sample_probability: 0.05
|
||
|
||
cross_streets:
|
||
and: *og
|
||
corner_of: &hjorne_av
|
||
canonical: hjørne av
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
at_the_corner_of: &pa_hjornet_av
|
||
canonical: på hjørnet av
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
intersection:
|
||
default: *og
|
||
probability: 0.7
|
||
alternatives:
|
||
- alternative: *hjorne_av
|
||
probability: 0.15
|
||
- alternative: *pa_hjornet_av
|
||
probability: 0.15
|
||
|
||
between:
|
||
canonical: mellom
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
parentheses_probability: 0.5
|
||
|
||
levels:
|
||
floor: &etasje
|
||
canonical: etasje
|
||
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
|
||
hovedetasje: &hovedetasje
|
||
canonical: hovedetasje
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric_affix:
|
||
affix: h
|
||
direction: left
|
||
zero_pad: 2
|
||
numeric_probability: 0.1
|
||
numeric_affix_probability: 0.9
|
||
underetasje: &underetasje
|
||
canonical: underetasje
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: u
|
||
direction: left
|
||
zero_pad: 2
|
||
numeric_probability: 0.1
|
||
numeric_affix_probability: 0.9
|
||
loftsetasje: &loftsetasje
|
||
canonical: loftsetasje
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: l
|
||
direction: left
|
||
zero_pad: 2
|
||
numeric_probability: 0.1
|
||
numeric_affix_probability: 0.9
|
||
loft: &loft
|
||
canonical: loft
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: right
|
||
kjeller: &kjeller
|
||
canonical: kjeller
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
# e.g. 1 kjeller
|
||
numeric:
|
||
direction: right
|
||
direction_probability: 0.8
|
||
# e.g. k1
|
||
numeric_affix:
|
||
affix: k
|
||
direction: left
|
||
# e.g. 1. k
|
||
ordinal:
|
||
direction: right
|
||
standalone_probability: 0.9
|
||
number_abs_value: true
|
||
number_min_abs_value: 1
|
||
numeric_probability: 0.005
|
||
numeric_affix_probability: 0.1
|
||
ordinal_probability: 0.005
|
||
aliases:
|
||
"<-1":
|
||
default: *kjeller
|
||
"-1":
|
||
default: *kjeller
|
||
probability: 0.9
|
||
alternatives:
|
||
- alternative: *etasje
|
||
probability: 0.1
|
||
"top":
|
||
default: *etasje
|
||
probability: 0.85
|
||
alternatives:
|
||
- alternative: *loftsetasje
|
||
probability: 0.1
|
||
- alternative: *loft
|
||
probability: 0.05
|
||
|
||
numbering_starts_at: 1
|
||
|
||
alphanumeric:
|
||
default: *etasje
|
||
probability: 0.95
|
||
alternatives:
|
||
- alternative: *underetasje
|
||
probability: 0.05
|
||
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: i nærheden af
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.7
|
||
alternatives:
|
||
- alternative:
|
||
canonical: tæt på
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.2
|
||
- alternative:
|
||
canonical: tæt ved
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
nearby:
|
||
default:
|
||
canonical: i nærheden
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.4
|
||
alternatives:
|
||
- alternative:
|
||
canonical: rundt her
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.2
|
||
- alternative:
|
||
canonical: nær her
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
- alternative:
|
||
canonical: nær
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
- alternative:
|
||
canonical: omkring her
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
- alternative:
|
||
canonical: tæt på her
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
near_me:
|
||
default:
|
||
canonical: nær mig
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.8
|
||
alternatives:
|
||
- alternative:
|
||
canonical: i nærheden af mig
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
- alternative:
|
||
canonical: tæt på mig
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
probability: 0.1
|
||
|
||
# Don't worry about agreement
|
||
in:
|
||
default:
|
||
canonical: i
|
||
probability: 0.8
|
||
alternatives:
|
||
- alternative:
|
||
canonical: om
|
||
probability: 0.1
|
||
- alternative:
|
||
canonical: på
|
||
probability: 0.1
|
||
|
||
|
||
# Probabilities of each phrase
|
||
near_probability: 0.35
|
||
nearby_probability: 0.2
|
||
near_me_probability: 0.1
|
||
in_probability: 0.35
|
||
|
||
|
||
directions:
|
||
right: &hoyre
|
||
canonical: høyre
|
||
sample: true
|
||
canonical_probability: 0.1
|
||
sample_probability: 0.9
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: h
|
||
direction: right
|
||
whitespace_probability: 0.1
|
||
numeric_probability: 0.8
|
||
numeric_affix_probability: 0.2
|
||
left: &igjen
|
||
canonical: igjen
|
||
sample: true
|
||
canonical_probability: 0.1
|
||
sample_probability: 0.9
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: i
|
||
direction: right
|
||
whitespace_probability: 0.1
|
||
numeric_probability: 0.8
|
||
numeric_affix_probability: 0.2
|
||
alternatives:
|
||
- alternative: *hoyre
|
||
probability: 0.5
|
||
- alternative: *igjen
|
||
probability: 0.5
|
||
|
||
|
||
cardinal_directions:
|
||
east: &ost
|
||
canonical: øst
|
||
abbreviated: ø
|
||
canonical_probability: 0.95
|
||
abbreviated_probability: 0.05
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: ø
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
west: &vest
|
||
canonical: vest
|
||
abbreviated: v
|
||
canonical_probability: 0.95
|
||
abbreviated_probability: 0.05
|
||
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: 0.95
|
||
abbreviated_probability: 0.05
|
||
numeric:
|
||
direction: right
|
||
numeric_affix:
|
||
affix: n
|
||
direction: right
|
||
numeric_probability: 0.5
|
||
numeric_affix_probability: 0.5
|
||
|
||
south: &syd
|
||
canonical: syd
|
||
abbreviated: s
|
||
sample: true
|
||
canonical_probability: 0.75
|
||
abbreviated_probability: 0.1
|
||
sample_probability: 0.15
|
||
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: *ost
|
||
probability: 0.25
|
||
- alternative: *syd
|
||
probability: 0.25
|
||
- alternative: *vest
|
||
probability: 0.25
|
||
|
||
|
||
entrances:
|
||
inngang: &inngang
|
||
canonical: inngang
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
|
||
# Eingang 1, Eingang A, etc.
|
||
alphanumeric: &entrance_alphanumeric
|
||
default: *inngang
|
||
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:
|
||
stiege: &stiege
|
||
canonical: stiege
|
||
abbreviated: stg
|
||
sample: true
|
||
canonical_probability: 0.7
|
||
abbreviated_probability: 0.2
|
||
sample_probability: 0.1
|
||
numeric:
|
||
direction: left
|
||
trapp: &trapp
|
||
canonical: trapp
|
||
sample: true
|
||
canonical_probability: 0.9
|
||
sample_probability: 0.1
|
||
numeric:
|
||
direction: left
|
||
|
||
|
||
alphanumeric: &staircase_alphanumeric
|
||
default: *trapp
|
||
probability: 0.8
|
||
alternatives:
|
||
- alternative: *stiege
|
||
probability: 0.2
|
||
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: *nord
|
||
- alternative: *syd
|
||
- alternative: *ost
|
||
- alternative: *vest
|
||
|
||
po_boxes:
|
||
postboks: &postboks
|
||
canonical: postboks
|
||
abbreviated: pb
|
||
sample: true
|
||
canonical_probability: 0.6
|
||
abbreviated_probability: 0.2
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.2 # Pb No 1234
|
||
boks: &boks
|
||
canonical: boks
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.2 # Boks No 1234
|
||
alphanumeric:
|
||
sample: false
|
||
default: *postboks
|
||
probability: 0.9
|
||
alternatives:
|
||
- alternative: *boks
|
||
probability: 0.1
|
||
numeric_probability: 0.9 # Pb 123
|
||
alpha_probability: 0.05 # Pb A
|
||
numeric_plus_alpha_probability: 0.04 # Pb 123G
|
||
alpha_plus_numeric_probability: 0.01 # Pb 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:
|
||
leilighet: &leilighet
|
||
canonical: leilighet
|
||
abbreviated: leil
|
||
sample: true
|
||
canonical_probability: 0.6
|
||
abbreviated_probability: 0.1
|
||
sample_probability: 0.3
|
||
numeric:
|
||
direction: left
|
||
null_phrase_probability: 0.4
|
||
# Lejlighed nummer 4
|
||
add_number_phrase: true
|
||
add_number_phrase_probability: 0.05
|
||
hus: &hus
|
||
canonical: hus
|
||
sample: true
|
||
canonical_probability: 0.8
|
||
sample_probability: 0.2
|
||
numeric:
|
||
direction: left
|
||
vaerelse: &vaerelse
|
||
canonical: værelse
|
||
sample: true
|
||
canonical_probability: 0.7
|
||
sample_probability: 0.3
|
||
numeric:
|
||
direction: left
|
||
alphanumeric: &unit_alphanumeric
|
||
default: *leilighet
|
||
probability: 0.8
|
||
alternatives:
|
||
- alternative: *hus
|
||
probability: 0.1
|
||
- alternative: *vaerelse
|
||
probability: 0.1
|
||
numeric_probability: 0.95 # e.g. Lejlighed 1
|
||
alpha_probability: 0.05 # e.g. Lejl A
|
||
|
||
alpha_plus_numeric:
|
||
whitespace_probability: 0.1
|
||
numeric_plus_alpha:
|
||
whitespace_probability: 0.1
|
||
|
||
# Separate random probability for adding directions like 2H, 2I, etc.
|
||
add_direction: true
|
||
add_direction_probability: 0.005
|
||
|
||
# Add directions for plain numbers
|
||
add_direction_numeric: true
|
||
# Add direction only e.g. Lejlighed Igjen
|
||
add_direction_standalone: true
|
||
|
||
# If there are 10 floors, create unit numbers like #301 or #1032
|
||
use_floor_probability: 0.2
|
||
|
||
# Use the actual floor phrase as long as the whole phrase is numeric
|
||
# Has the effect of creating Bolignummer-style units
|
||
use_floor_numeric_affix_probability: 0.7
|
||
use_floor_affix_floor_num_digits: 2
|
||
use_floor_affix_unit_num_digits: 2
|