feat: add middle initial and new venue classification (#117)

classification/MiddleInitialClassification.js

@@ -0,0 +1,10 @@
+const Classification = require('../classification/Classification')
+
+class MiddleInitialClassification extends Classification {
+  constructor (confidence, meta) {
+    super(confidence, meta)
+    this.label = 'middle_initial'
+  }
+}
+
+module.exports = MiddleInitialClassification

classification/MiddleInitialClassification.test.js

@@ -0,0 +1,24 @@
+const Classification = require('./MiddleInitialClassification')
+
+module.exports.tests = {}
+
+module.exports.tests.constructor = (test) => {
+  test('constructor', (t) => {
+    let c = new Classification()
+    t.false(c.public)
+    t.equals(c.label, 'middle_initial')
+    t.equals(c.confidence, 1.0)
+    t.deepEqual(c.meta, {})
+    t.end()
+  })
+}
+
+module.exports.all = (tape, common) => {
+  function test (name, testFunction) {
+    return tape(`MiddleInitialClassification: ${name}`, testFunction)
+  }
+
+  for (var testCase in module.exports.tests) {
+    module.exports.tests[testCase](test, common)
+  }
+}

classifier/MiddleInitialClassifier.js

@@ -0,0 +1,17 @@
+const PhraseClassifier = require('./super/PhraseClassifier')
+const MiddleInitialClassification = require('../classification/MiddleInitialClassification')
+
+const SingleLetterRegExp = /^[A-Za-z]\.?$/
+
+class MiddleInitialClassifier extends PhraseClassifier {
+  setup () {
+  }
+
+  each (span) {
+    if (SingleLetterRegExp.test(span.body)) {
+      span.classify(new MiddleInitialClassification(1))
+    }
+  }
+}
+
+module.exports = MiddleInitialClassifier

classifier/MiddleInitialClassifier.test.js

@@ -0,0 +1,56 @@
+const MiddleInitialClassifier = require('./MiddleInitialClassifier')
+const MiddleInitialClassification = require('../classification/MiddleInitialClassification')
+const Span = require('../tokenization/Span')
+const classifier = new MiddleInitialClassifier()
+
+module.exports.tests = {}
+
+function classify (body) {
+  let s = new Span(body)
+  classifier.each(s, null, 1)
+  return s
+}
+
+module.exports.tests.classify = (test) => {
+  const valid = [
+    'M.',
+    'M'
+  ]
+
+  valid.forEach(token => {
+    test(`classify: ${token}`, (t) => {
+      let s = classify(token)
+      t.deepEqual(s.classifications, {
+        MiddleInitialClassification: new MiddleInitialClassification(1.0)
+      })
+      t.end()
+    })
+  })
+
+  const invalid = [
+    'Mae',
+    '122',
+    'M,',
+    '&',
+    'Mr',
+    'Esq'
+  ]
+
+  invalid.forEach(token => {
+    test(`classify: ${token}`, (t) => {
+      let s = classify(token)
+      t.deepEqual(s.classifications, {})
+      t.end()
+    })
+  })
+}
+
+module.exports.all = (tape, common) => {
+  function test (name, testFunction) {
+    return tape(`MiddleInitialClassifier: ${name}`, testFunction)
+  }
+
+  for (var testCase in module.exports.tests) {
+    module.exports.tests[testCase](test, common)
+  }
+}

classifier/scheme/person.js

@@ -53,6 +53,7 @@ module.exports = [
   },
   {
     // Raul Leite Magalhães (first name, middle name, family name)
+    // Donald W. Reynolds
     confidence: 0.5,
     Class: PersonClassification,
     scheme: [
@@ -61,7 +62,7 @@ module.exports = [
         not: ['StreetClassification', 'IntersectionClassification']
       },
       {
-        is: ['GivenNameClassification', 'SurnameClassification'],
+        is: ['GivenNameClassification', 'SurnameClassification', 'MiddleInitialClassification'],
         not: ['StreetClassification', 'IntersectionClassification']
       },
       {

classifier/scheme/venue.js

@@ -129,6 +129,22 @@ module.exports = [
       }
     ]
   },
+  {
+    // Donald W Reynolds Stadium
+    // boost confidence slightly above street for "Donald W"
+    confidence: 0.82,
+    Class: VenueClassification,
+    scheme: [
+      {
+        is: ['PersonClassification'],
+        not: ['StreetClassification']
+      },
+      {
+        is: ['PlaceClassification', 'VenueClassification'],
+        not: ['StreetClassification']
+      }
+    ]
+  },
   {
     // ZAC du Pré
     confidence: 0.8,

parser/AddressParser.js

@@ -17,6 +17,7 @@ const StopWordClassifier = require('../classifier/StopWordClassifier')
 const PersonClassifier = require('../classifier/PersonClassifier')
 const GivenNameClassifier = require('../classifier/GivenNameClassifier')
 const SurnameClassifier = require('../classifier/SurnameClassifier')
+const MiddleInitialClassifier = require('../classifier/MiddleInitialClassifier')
 const PersonalSuffixClassifier = require('../classifier/PersonalSuffixClassifier')
 const PersonalTitleClassifier = require('../classifier/PersonalTitleClassifier')
 const ChainClassifier = require('../classifier/ChainClassifier')
@@ -67,6 +68,7 @@ class AddressParser extends Parser {
         new PersonClassifier(),
         new GivenNameClassifier(),
         new SurnameClassifier(),
+        new MiddleInitialClassifier(),
         new PersonalSuffixClassifier(),
         new PersonalTitleClassifier(),
         new ChainClassifier(),

test/venue.usa.test.js

@@ -11,6 +11,16 @@ const testcase = (test, common) => {
     { locality: 'NYC' }
   ])
 
+  // checking that NAME INITIAL NAME works
+  assert('Donald W Reynolds Stadium', [
+    { venue: 'Donald W Reynolds Stadium' }
+  ])
+
+  assert('Donald W. Reynolds Stadium', [
+    { venue: 'Donald W. Reynolds Stadium' }
+  ])
+
+  // checking that "japan" is interpreted as a country, not a city
   assert('Universal Studios Japan', [
     { venue: 'Universal Studios' },
     { country: 'Japan' }