440 lines
12 KiB
YAML
440 lines
12 KiB
YAML
# hu.yaml
|
|
# -------
|
|
# Hungarian language specification.
|
|
|
|
components:
|
|
level:
|
|
null_probability: 0.75
|
|
alphanumeric_probability: 0.2
|
|
standalone_probability: 0.05
|
|
|
|
unit:
|
|
null_probability: 0.75
|
|
alphanumeric_probability: 0.25
|
|
|
|
combinations:
|
|
-
|
|
components:
|
|
- level
|
|
- unit
|
|
label: unit
|
|
separators:
|
|
- separator: "/"
|
|
probability: 0.55
|
|
- separator: " "
|
|
probability: 0.4
|
|
- separator: "-"
|
|
probability: 0.05
|
|
probability: 0.8
|
|
|
|
|
|
numbers:
|
|
default: &szam
|
|
canonical: szám
|
|
sample: true
|
|
# Probabilities
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
numeric_probability: 0.4
|
|
numeric_affix_probability: 0.6
|
|
|
|
and:
|
|
default: &es
|
|
canonical: és
|
|
abbreviated: "&"
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.75
|
|
sample: true
|
|
sample_probability: 0.05
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: &es_a
|
|
canonical: és a
|
|
canonical_probability: 0.9
|
|
sample: true
|
|
sample_probability: 0.1
|
|
probability: 0.2
|
|
- alternative: &es_az
|
|
canonical: és az
|
|
canonical_probability: 0.9
|
|
sample: true
|
|
sample_probability: 0.1
|
|
probability: 0.2
|
|
|
|
cross_streets:
|
|
and: *es
|
|
corner_of: &sarkan
|
|
canonical: sarkán
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
intersection:
|
|
default: *es
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: *es_a
|
|
probability: 0.1
|
|
- alternative: *es_az
|
|
probability: 0.1
|
|
- alternative: *sarkan
|
|
probability: 0.2
|
|
|
|
between:
|
|
canonical: között
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
parentheses_probability: 0.5
|
|
|
|
levels:
|
|
floor: &emelet
|
|
canonical: emelet
|
|
abbreviated: em
|
|
sample: true
|
|
canonical_probability: 0.1
|
|
abbreviated_probability: 0.85
|
|
sample_probability: 0.05
|
|
numeric:
|
|
direction: left
|
|
direction_probability: 0.9
|
|
ordinal:
|
|
direction: right
|
|
digits:
|
|
ascii_probability: 0.2
|
|
roman_numeral_probability: 0.8
|
|
numeric_probability: 0.1
|
|
ordinal_probability: 0.9
|
|
foldszint: &foldszint
|
|
canonical: földszint
|
|
abbreviated: fszt
|
|
sample: true
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.6
|
|
sample_probability: 0.2
|
|
felemelet: &felemelet
|
|
canonical: félemelet
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
magasfoldszint: &magasfoldszint
|
|
canonical: magasföldszint
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
pince: &pince
|
|
canonical: pince
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
standalone_probability: 0.99
|
|
number_abs_value: true
|
|
number_min_abs_value: 1
|
|
numeric_probability: 0.005
|
|
ordinal_probability: 0.005
|
|
alagsor: &alagsor
|
|
canonical: alagsor
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
standalone_probability: 0.99
|
|
number_abs_value: true
|
|
number_min_abs_value: 1
|
|
numeric_probability: 0.005
|
|
ordinal_probability: 0.005
|
|
felszuteren: &felszuteren
|
|
canonical: félszuterén
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
standalone_probability: 0.99
|
|
number_abs_value: true
|
|
number_min_abs_value: 1
|
|
numeric_probability: 0.005
|
|
ordinal_probability: 0.005
|
|
szuteren: &szuteren
|
|
canonical: szuterén
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
standalone_probability: 0.99
|
|
number_abs_value: true
|
|
number_min_abs_value: 1
|
|
numeric_probability: 0.005
|
|
ordinal_probability: 0.005
|
|
aliases:
|
|
"<-1":
|
|
default: *alagsor
|
|
probability: 0.6
|
|
alternatives:
|
|
- alternative: *pince
|
|
probability: 0.3
|
|
- alternative: *szuteren
|
|
probability: 0.1
|
|
"-1":
|
|
default: *alagsor
|
|
probability: 0.5
|
|
alternatives:
|
|
- alternative: *pince
|
|
probability: 0.3
|
|
- alternative: *szuteren
|
|
probability: 0.1
|
|
- alternative: *felszuteren
|
|
probability: 0.1
|
|
|
|
"0":
|
|
default: *foldszint
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *emelet
|
|
probability: 0.1
|
|
|
|
"1":
|
|
default: *emelet
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *felemelet
|
|
probability: 0.1
|
|
|
|
"2":
|
|
default: *emelet
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *magasfoldszint
|
|
probability: 0.1
|
|
|
|
numbering_starts_at: 0
|
|
|
|
alphanumeric:
|
|
default: *emelet
|
|
numeric_probability: 0.59 # With this probability, pick an integer
|
|
roman_numeral_probability: 0.4 # Pick a Roman numeral for the actual value
|
|
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: közelében
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
nearby:
|
|
default:
|
|
canonical: közelben
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
|
|
near_me:
|
|
default:
|
|
canonical: közelemben
|
|
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: &jobb
|
|
canonical: jobb
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
left: &bal
|
|
canonical: bal
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: right
|
|
alternatives:
|
|
- alternative: *jobb
|
|
probability: 0.5
|
|
- alternative: *bal
|
|
probability: 0.5
|
|
|
|
cardinal_directions:
|
|
east: &kelet
|
|
canonical: kelet
|
|
abbreviated: k
|
|
canonical_probability: 0.95
|
|
abbreviated_probability: 0.05
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: k
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
west: &nyugat
|
|
canonical: nyugat
|
|
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
|
|
|
|
north: &eszak
|
|
canonical: észak
|
|
abbreviated: e
|
|
canonical_probability: 0.95
|
|
abbreviated_probability: 0.05
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: e
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
south: &del
|
|
canonical: dél
|
|
abbreviated: d
|
|
sample: true
|
|
canonical_probability: 0.75
|
|
abbreviated_probability: 0.1
|
|
sample_probability: 0.15
|
|
numeric:
|
|
direction: right
|
|
numeric_affix:
|
|
affix: d
|
|
direction: right
|
|
numeric_probability: 0.5
|
|
numeric_affix_probability: 0.5
|
|
|
|
alternatives:
|
|
- alternative: *eszak
|
|
probability: 0.25
|
|
- alternative: *kelet
|
|
probability: 0.25
|
|
- alternative: *del
|
|
probability: 0.25
|
|
- alternative: *nyugat
|
|
probability: 0.25
|
|
|
|
|
|
po_boxes:
|
|
postafiok: &postafiok
|
|
canonical: postafiók
|
|
abbreviated: pf
|
|
sample: true
|
|
canonical_probability: 0.2
|
|
abbreviated_probability: 0.7
|
|
sample_probability: 0.1
|
|
numeric:
|
|
direction: left
|
|
alphanumeric:
|
|
default: *postafiok
|
|
numeric_probability: 0.9 # Pf 123
|
|
alpha_probability: 0.05 # Pf A
|
|
numeric_plus_alpha_probability: 0.04 # Pf 123G
|
|
alpha_plus_numeric_probability: 0.01 # Pf 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:
|
|
lakas: &lakas
|
|
canonical: lakás
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
ordinal:
|
|
direction: right
|
|
numeric_probability: 0.3
|
|
ordinal_probability: 0.7
|
|
iroda: &iroda
|
|
canonical: iroda
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
szoba: &szoba
|
|
canonical: szoba
|
|
sample: true
|
|
canonical_probability: 0.8
|
|
sample_probability: 0.2
|
|
numeric:
|
|
direction: left
|
|
|
|
alphanumeric: &unit_alphanumeric
|
|
default: *lakas
|
|
probability: 0.9
|
|
alternatives:
|
|
- alternative: *szoba
|
|
probability: 0.1
|
|
numeric_probability: 0.95 # e.g. m. 1
|
|
numeric_plus_alpha_probability: 0.005 # e.g. 1A
|
|
alpha_plus_numeric_probability: 0.005 # e.g. A1
|
|
alpha_probability: 0.04 # e.g. m. A
|
|
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
|
|
# If there are 10 floors, create unit numbers like #301 or #1032
|
|
use_floor_probability: 0.2
|
|
|
|
zones:
|
|
commercial: &commercial_unit_types
|
|
default: *iroda
|
|
numeric_probability: 0.95 # e.g. pokój 1
|
|
numeric_plus_alpha_probability: 0.01 # e.g. pokój 1A
|
|
alpha_plus_numeric_probability: 0.01 # e.g. pokój A1
|
|
alpha_probability: 0.03 # e.g. pokój A
|
|
alpha_plus_numeric:
|
|
whitespace_probability: 0.1
|
|
numeric_plus_alpha:
|
|
whitespace_probability: 0.1
|
|
university: *commercial_unit_types
|