tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[addresses] Adding more numeric/numeric_affix probabilities to English config

Browse Source
This commit is contained in:
Al
2016-04-19 11:25:12 -04:00
parent ce2b2d9559
commit c31926f3dd
1 changed files with 109 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

134
resources/addresses/en.yaml
View File

@@ -631,33 +631,58 @@ directions:
abbreviated: r
canonical_probability: 0.7
abbreviated_probability: 0.3
numeric:
direction: right
numeric_affix:
affix: r
direction: right
numeric_probability: 0.2
numeric_affix_probability: 0.8
left: &left
canonical: left
abbreviated: l
canonical_probability: 0.7
abbreviated_probability: 0.3
numeric:
direction: right
numeric_affix:
affix: l
direction: right
numeric_probability: 0.2
numeric_affix_probability: 0.8
rear: &rear
canonical: rear
abbreviated: r
canonical_probability: 0.8
abbreviated_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: r
direction: right
numeric_probability: 0.2
numeric_affix_probability: 0.8
front: &front
canonical: front
abbreviated: frnt
canonical_probability: 0.8
abbreviated_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: f
direction: right
numeric_probability: 0.2
numeric_affix_probability: 0.8
alternatives:
- alternative: *right
probability: 0.45
- alternative: *left
probability: 0.45
- alternative: *front
probability: 0.05
- alternative: *rear
probability: 0.05
cardinal_directions:
east: &east
@@ -667,9 +692,13 @@ cardinal_directions:
canonical_probability: 0.5
abbreviated_probability: 0.3
sample_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: e
direction: right
numeric_probability: 0.6
numeric_affix_probability: 0.4
west: &west
canonical: west
abbreviated: w
@@ -677,9 +706,13 @@ cardinal_directions:
canonical_probability: 0.5
abbreviated_probability: 0.3
sample_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: w
direction: right
numeric_probability: 0.6
numeric_affix_probability: 0.4
north: &north
canonical: north
abbreviated: n
@@ -687,9 +720,13 @@ cardinal_directions:
canonical_probability: 0.5
abbreviated_probability: 0.3
sample_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: n
direction: right
numeric_probability: 0.6
numeric_affix_probability: 0.4
south: &south
canonical: south
abbreviated: s
@@ -697,9 +734,23 @@ cardinal_directions:
canonical_probability: 0.5
abbreviated_probability: 0.3
sample_probability: 0.2
numeric:
direction: right
numeric_affix:
affix: s
direction: right
numeric_probability: 0.6
numeric_affix_probability: 0.4
alternatives:
- alternative: *north
probability: 0.25
- alternative: *east
probability: 0.25
- alternative: *south
probability: 0.25
- alternative: *west
probability: 0.25
# Entrance
# ========
@@ -819,6 +870,8 @@ units:
abbreviated: stes
canonical_probability: 0.6
abbreviated_probability: 0.4
numeric:
direction: left
# Suite #101 and Suite No. 101 as opposed to Suite 101
add_number_phrase: true
add_number_phrase_probability: 0.5
@@ -831,6 +884,10 @@ units:
sample_probability: 0.2
plural:
canonical: penthouses
numeric:
direction: left
numeric_probability: 0.2
standalone_probability: 0.8
# Penthouse #1 and Penthouse No. 1
add_number_phrase: true
add_number_phrase_probability: 0.2
@@ -874,6 +931,8 @@ units:
abbreviated: ofcs
canonical_probability: 0.4
abbreviated_probability: 0.6
numeric:
direction: left
# Office #1 and Office No. 1
add_number_phrase: true
add_number_phrase_probability: 0.7
@@ -884,6 +943,8 @@ units:
sample_probability: 0.2
plural:
canonical: doors
numeric:
direction: left
# Door #1 and Door No. 1
add_number_phrase: true
add_number_phrase_probability: 0.2
@@ -898,6 +959,8 @@ units:
abbreviated: rms
canonical_probability: 0.6
abbreviated_probability: 0.4
numeric:
direction: left
# Room #1 and Room No. 1
add_number_phrase: true
add_number_phrase_probability: 0.6
@@ -905,6 +968,11 @@ units:
canonical: hall
plural:
canonical: halls
numeric:
direction: left
# Room #1 and Room No. 1
add_number_phrase: true
add_number_phrase_probability: 0.6
apartment: &apartment
canonical: apartment
abbreviated: apt
@@ -918,6 +986,8 @@ units:
abbreviated: apts
canonical_probability: 0.2
abbreviated: 0.8
numeric:
direction: left
# Apt #1 and Apt No. 1
add_number_phrase: true
add_number_phrase_probability: 0.4
@@ -926,13 +996,15 @@ units:
abbreviated: flt
sample: true
canonical_probability: 0.8
abbreviated_probability: 0.1
sample_probability: 0.1
abbreviated_probability: 0.15
sample_probability: 0.05
plural:
canonical: flats
abbreviated: flts
canonical_probability: 0.6
abbreviated_probability: 0.4
canonical_probability: 0.8
abbreviated_probability: 0.2
numeric:
direction: left
# Flat #1 and Flat No. 1
add_number_phrase: true
add_number_phrase_probability: 0.4
@@ -943,16 +1015,20 @@ units:
sample_probability: 0.1
plural:
canonical: lots
numeric:
direction: left
# Lot #1 and Lot No. 1
add_number_phrase: true
add_number_phrase_probability: 0.6
parcel: &parcel
canonical: parcel
canonical_probability: 0.9
sample: true
canonical_probability: 0.9
sample_probability: 0.1
plural:
canonical: parcels
numeric:
direction: left
add_number_phrase: true
add_number_phrase_probability: 0.6
unit: &unit
@@ -964,6 +1040,8 @@ units:
sample_probability: 0.1
plural:
canonical: units
numeric:
direction: left
# Unit #1 and Unit No. 1
add_number_phrase: true
add_number_phrase_probability: 0.4
@@ -1007,35 +1085,36 @@ units:
probability: 0.2
industrial:
default: *lot
probability: 0.7
probability: 0.5
alternatives:
- alternative: *suite
probability: 0.15
probability: 0.3
- alternative: *unit
probability: 0.1
probability: 0.19
- alternative: *parcel
alternatives: 0.05
probability: 0.01
university:
default: *room
probability: 0.9
alternatives:
- alternative: *hall
probability: 0.1
allotments:
lot:
default: *lot
numeric_probability: 0.8
alphanumeric_probability: 0.1
alpha_probability: 0.1
parcel:
default: *parcel
numeric_probability: 0.3
alphanumeric_probability: 0.3
alpha_probability: 0.4
lot_probability: 0.9
parcel_probability: 0.06
lot_plus_parcel_probability: 0.02
parcel_plus_lot_probability: 0.02
allotments:
lot:
default: *lot
numeric_probability: 0.8
alphanumeric_probability: 0.1
alpha_probability: 0.1
parcel:
default: *parcel
numeric_probability: 0.3
alphanumeric_probability: 0.3
alpha_probability: 0.4
lot_probability: 0.9
parcel_probability: 0.06
lot_plus_parcel_probability: 0.02
parcel_plus_lot_probability: 0.02
directional:
modifier:
@@ -1212,14 +1291,17 @@ countries:
sample: true
default: *flat
# Reduce the default's probability to make room for sampling
probability: 0.59
probability: 0.39
sample_probability: 0.01
numeric:
direction: left
standalone:
default: *penthouse
sample: true
# Reduce the default's probability to make room for sampling
probability: 0.39
sample_probability: 0.01
standalone_probability: 1.0
# For unit types like 2/34
combined:
@@ -1240,3 +1322,5 @@ countries:
nz:
po_boxes: *australia_po_boxes
units: *australia_unit_types