more place names

data/lexicon/nouns/properNouns.js

@@ -15,6 +15,27 @@ export default [
   'stevens',
   'abid',
   'mcgill',
+  'hudson',
+  'chesley',
+  'carling',
+  'berkeley',
+  'beeton',
+  'carleton',
+  'ajax',
+  'weston',
+  'sherwood',
+  'wembley',
+  'hinton',
+  'bentley',
+  'landsdowne',
+  'brock',
+  'dalhousie',
+  'spalding',
+  'charlton',
+  'rothwell',
+  'gosford',
+  'frampton',
+  'fairview',
   // currencies
   'nis',
   'riel',

data/lexicon/nouns/singulars.js

@@ -431,4 +431,5 @@ export default [
   'cannibal',
   'video game',
   'layer',
+  'canton',
 ]

data/lexicon/places/places.js

@@ -133,4 +133,192 @@ export default [
   'cedar breaks',
   'scotts bluff',
   'saint lawrence river',
+
+  // generic place names
+  'west end',
+  'east end',
+  'north end',
+  'south end',
+  'west side',
+  'east side',
+  'north side',
+  'south side',
+  'centerville',
+  'clearwater',
+  'Maplewood',
+  'Pinecrest',
+  'Brookside',
+  'Riverside',
+  'Willowbrook',
+
+  //misc notable british
+  'abbotsford',
+  'abingdon',
+  'acton',
+  'agincourt',
+  'alberton',
+  'aldershot',
+  'alliston',
+  'ancaster',
+  'andover',
+  'ashcroft',
+  'aylesford',
+  'baltimore',
+  'barnsley',
+  'battersea',
+  'beaconsfield',
+  'bedford',
+  'bellingham',
+  'berwick',
+  'bexhill',
+  'bingley',
+  'blackburn',
+  'blackpool',
+  'blackwall',
+  'bolton',
+  'bracebridge',
+  'bramley',
+  'brampton',
+  'bridgenorth',
+  'bridgetown',
+  'brigham',
+  'brighton',
+  'brockton',
+  'buckingham',
+  'caldwell',
+  'cambridge',
+  'campbellford',
+  'carlisle',
+  'caversham',
+  'charlottetown',
+  'chelmsford',
+  'cheltenham',
+  'colchester',
+  'collingwood',
+  'compton',
+  'cornwall',
+  'croydon',
+  'darlington',
+  'dartford',
+  'dartmoor',
+  'denham',
+  'derry',
+  'devonport',
+  'digby',
+  'doncaster',
+  'dorchester',
+  'dover',
+  'dryden',
+  'dumfries',
+  'dunstable',
+  'durham',
+  'dutton',
+  'earlscourt',
+  'eglinton',
+  'enfield',
+  'epping',
+  'euston',
+  'everton',
+  'exeter',
+  'exford',
+  'gravenhurst',
+  'grimsby',
+  'guelph',
+  'haliburton',
+  'hampstead',
+  'hampton',
+  'harford',
+  'harrow',
+  'hastings',
+  'hereford',
+  'hertford',
+  'heywood',
+  'highbury',
+  'huntingdon',
+  'ilford',
+  'inglewood',
+  'inverness',
+  'ipswich',
+  'kensington',
+  'keswick',
+  'kitchener',
+  'lambeth',
+  'lancaster',
+  'langford',
+  'langley',
+  'langton',
+  'lansing',
+  'leaside',
+  'lewes',
+  'lindley',
+  'lyndhurst',
+  'maidstone',
+  'malton',
+  'malvern',
+  'mansfield',
+  'markham',
+  'millington',
+  'newark',
+  'newham',
+  'newmarket',
+  'newport',
+  'northampton',
+  'norwich',
+  'nottingham',
+  'orford',
+  'oxford',
+  'oxley',
+  'parkwood',
+  'pelham',
+  'pembroke',
+  'pennington',
+  'penzance',
+  'perth',
+  'pickering',
+  'picton',
+  'putnam',
+  'rayleigh',
+  'redcliffe',
+  'redhill',
+  'richmond',
+  'rochester',
+  'seaforth',
+  'seaton',
+  'sherbrooke',
+  'southampton',
+  'southwold',
+  'stafford',
+  'stockwell',
+  'stratford',
+  'stroud',
+  'sudbury',
+  'sunderland',
+  'sydenham',
+  'taunton',
+  'thornhill',
+  'tiverton',
+  'torquay',
+  'tottenham',
+  'truro',
+  'upton',
+  'uxbridge',
+  'vauxhall',
+  'wakefield',
+  'warwick',
+  'waterford',
+  'watford',
+  'weedon',
+  'wentworth',
+  'whitby',
+  'whitchurch',
+  'wickham',
+  'winchester',
+  'wingham',
+  'woking',
+  'woodbridge',
+  'woodside',
+  'woodstock',
+  'woolwich',
+  'wycombe',
+  'york',
 ]

data/lexicon/places/regions.js

@@ -223,4 +223,14 @@ export default [
   'chechnya',
   'chuvashia',
   'crimea',
+
+  // malaysia
+  'penang',
+  'pahang',
+  'perak',
+  'sarawak',
+  'selangor',
+  'malacca',
+  'kedah',
+  'kelantan',
 ]

scratch.js

@@ -4,7 +4,119 @@ import nlp from './src/three.js'
 // nlp.plugin(plg)
 // nlp.verbose('tagger')
 
-let doc = nlp('john smith, judy cline').people().debug()
+let doc = nlp(` Maplewood
+Pinecrest
+Oakville
+Brookside
+Springdale
+Riverside
+Willowbrook
+Meadowview
+Cedar Falls
+Lakeview
+Sunset Hills
+Valley Springs
+Mountainview
+Greenfield
+Parkville
+Fairview
+Woodland Heights
+Harbor Town
+Highland Park
+Clearwater Junction
+Maple Grove
+Pine Ridge
+Oak Valley
+Brookhaven
+Spring Valley
+Riverdale
+Willow Springs
+Meadowbrook
+Cedar Ridge
+Lakeside
+Sunset Terrace
+Valley View
+Mountain Valley
+Green Hills
+Parkside
+Fair Oaks
+Woodland Hills
+Harbor Springs
+Highland Village
+Clear Creek
+Maple Ridge
+Pine Lake
+Oakwood
+Brookfield
+Spring Creek
+Riverbank
+Willow Creek
+Meadowland
+Cedar Creek
+Lake Shore
+Sunset Ridge
+Valley Ridge
+Mountain Top
+Green Valley
+Park View
+Fairbanks
+Woodland Park
+Harbor Point
+Highlands
+Clearview
+Maple Lane
+Pine Hill
+Oakdale
+Brooks Crossing
+Springhill
+Riverside Park
+Willow Grove
+Meadowlands
+Cedar Springs
+Lakeside Estates
+Sunset View
+Valley Heights
+Mountain Vista
+Green Acres
+Park Lane
+Fairmont
+Woodside
+Harbor Heights
+High Point
+Clearbrook
+Maple Falls
+Pinecrest Heights
+Oakridge
+Brookshire
+Springville
+Riverwood
+Willow Lake
+Meadow Heights
+Cedar Grove
+Lakeview Terrace
+Sunset Place
+Valley Park
+Mountain Meadow
+Greenfield Estates
+Park Place
+Fairfield
+Woodridge
+Harbor Village
+Hillcrest
+Clearview Heights
+
+`)
+  .ifNo('#Place')
+  .debug()
+
+// -bury
+// -ford
+// -ton
+// -shire
+
+// Dundas Harbour
+// hants county
+
 // console.log(nlp(`IEEE/WIC`).debug().docs[0][0])
 
 // const prependingText = 'the patient will need an '

src/2-two/postTagger/model/nouns/places.js

@@ -25,6 +25,8 @@ export default [
     tag: 'Address',
     reason: 'address-st',
   },
+  // port dover
+  { match: '(port|mount) #ProperName', tag: 'Place', reason: 'port-name' },
   // Sports Arenas and Complexs
   // {
   //   match: