de-compromise 0.0.9 → 0.1.0

package/src/01-one/tagset/tags/nouns.js

@@ -0,0 +1,116 @@
+const entity = ['Person', 'Place', 'Organization']
+
+export default {
+  Noun: {
+    not: ['Verb', 'Adjective', 'Adverb', 'Value', 'Determiner'],
+  },
+  Singular: {
+    is: 'Noun',
+    not: ['Plural'],
+  },
+  ProperNoun: {
+    is: 'Noun',
+  },
+  Person: {
+    is: 'Singular',
+    also: ['ProperNoun'],
+    not: ['Place', 'Organization', 'Date'],
+  },
+  FirstName: {
+    is: 'Person',
+  },
+  MaleName: {
+    is: 'FirstName',
+    not: ['FemaleName', 'LastName'],
+  },
+  FemaleName: {
+    is: 'FirstName',
+    not: ['MaleName', 'LastName'],
+  },
+  LastName: {
+    is: 'Person',
+    not: ['FirstName'],
+  },
+  Honorific: {
+    is: 'Noun',
+    not: ['FirstName', 'LastName', 'Value'],
+  },
+  Place: {
+    is: 'Singular',
+    not: ['Person', 'Organization'],
+  },
+  Country: {
+    is: 'Place',
+    also: ['ProperNoun'],
+    not: ['City'],
+  },
+  City: {
+    is: 'Place',
+    also: ['ProperNoun'],
+    not: ['Country'],
+  },
+  Region: {
+    is: 'Place',
+    also: ['ProperNoun'],
+  },
+  Address: {
+    // is: 'Place',
+  },
+  Organization: {
+    is: 'ProperNoun',
+    not: ['Person', 'Place'],
+  },
+  SportsTeam: {
+    is: 'Organization',
+  },
+  School: {
+    is: 'Organization',
+  },
+  Company: {
+    is: 'Organization',
+  },
+  Plural: {
+    is: 'Noun',
+    not: ['Singular'],
+  },
+  Uncountable: {
+    is: 'Noun',
+  },
+  Pronoun: {
+    is: 'Noun',
+    not: entity,
+  },
+  Actor: {
+    is: 'Noun',
+    not: entity,
+  },
+  Activity: {
+    is: 'Noun',
+    not: ['Person', 'Place'],
+  },
+  Unit: {
+    is: 'Noun',
+    not: entity,
+  },
+  Demonym: {
+    is: 'Noun',
+    also: ['ProperNoun'],
+    not: entity,
+  },
+  // standalone — 'meinem' is #Possessive as both a determiner ('meinem Freund')
+  // and a pronoun ('das ist meins')
+  Possessive: {},
+  // german genders
+  MaleNoun: {
+    is: 'Noun',
+    not: ['FemaleNoun', 'NeuterNoun'],
+  },
+  FemaleNoun: {
+    is: 'Noun',
+    not: ['MaleNoun', 'NeuterNoun'],
+  },
+  NeuterNoun: {
+    is: 'Noun',
+    not: ['MaleNoun', 'FemaleNoun'],
+  },
+}

package/src/01-one/tagset/tags/values.js

@@ -0,0 +1,47 @@
+export default {
+  Value: {
+    not: ['Verb', 'Adjective', 'Adverb'],
+  },
+  Ordinal: {
+    is: 'Value',
+    not: ['Cardinal'],
+  },
+  Cardinal: {
+    is: 'Value',
+    not: ['Ordinal'],
+  },
+  Fraction: {
+    is: 'Value',
+    not: ['Noun'],
+  },
+  Multiple: {
+    is: 'Value',
+  },
+  RomanNumeral: {
+    is: 'Cardinal',
+    not: ['TextValue'],
+  },
+  TextValue: {
+    is: 'Value',
+    not: ['NumericValue'],
+  },
+  TextOrdinal: {
+    is: 'TextValue',
+    also: ['Ordinal']
+  },
+  TextCardinal: {
+    is: 'TextValue',
+    also: ['Cardinal']
+  },
+
+  NumericValue: {
+    is: 'Value',
+    not: ['TextValue'],
+  },
+  Money: {
+    is: 'Cardinal',
+  },
+  Percent: {
+    is: 'Value',
+  },
+}

package/src/01-one/tagset/tags/verbs.js

@@ -0,0 +1,64 @@
+export default {
+  Verb: {
+    not: ['Noun', 'Adjective', 'Adverb', 'Value', 'Expression'],
+  },
+  PresentTense: {
+    is: 'Verb',
+    not: ['PastTense'],
+  },
+  Infinitive: {
+    is: 'PresentTense',
+    not: ['Gerund'],
+  },
+  Imperative: {
+    is: 'Infinitive',
+  },
+  Subjunctive: {
+    is: 'Verb',
+  },
+  Participle: {
+    is: 'Verb',
+  },
+  Gerund: {
+    is: 'PresentTense',
+    not: ['Copula'],
+  },
+  PastTense: {
+    is: 'Verb',
+    not: ['PresentTense', 'Gerund'],
+  },
+  Copula: {
+    is: 'Verb',
+  },
+  Modal: {
+    is: 'Verb',
+    not: ['Infinitive'],
+  },
+  PerfectTense: {
+    is: 'Verb',
+    not: ['Gerund'],
+  },
+  Pluperfect: {
+    is: 'Verb',
+  },
+  PhrasalVerb: {
+    is: 'Verb',
+  },
+  Particle: {
+    is: 'PhrasalVerb',
+    not: ['PastTense', 'PresentTense', 'Copula', 'Gerund'],
+  },
+  Auxiliary: {
+    is: 'Verb',
+    not: ['PastTense', 'PresentTense', 'Gerund', 'Conjunction'],
+  },
+  FirstPerson: {
+    is: 'Verb'
+  },
+  SecondPerson: {
+    is: 'Verb'
+  },
+  ThirdPerson: {
+    is: 'Verb'
+  },
+}

package/src/01-one/tokenizer/contractions.js

@@ -0,0 +1,40 @@
+// https://www.fluentu.com/blog/german/german-contractions/
+export default [
+  { word: 'am', out: ['an', 'dem'] },  //(on the / at the)
+  { word: 'ans', out: ['an', 'das'] },  //(on the / at the)
+  { word: 'aufs', out: ['auf', 'das'] },  //(on the / at the)
+  { word: 'beim', out: ['bei', 'dem'] },  //(at the / in the / with the)
+  { word: 'durchs', out: ['durch', 'das'] },  //(through the)
+  { word: 'fürs', out: ['für', 'das'] },  //(for the)
+  { word: 'hinterm', out: ['hinter', 'dem'] },  //(behind the)
+  { word: 'ins', out: ['in', 'das'] },  //(in the / into the / to the)
+  { word: 'im', out: ['in', 'dem'] },  //(at / in the)
+  { word: 'übers', out: ['über', 'das'] },  //(over the / about the)
+  { word: 'ums', out: ['um', 'das'] },  //(at the / around)
+  { word: 'unters', out: ['unter', 'das'] },  //(under the)
+  { word: 'unterm', out: ['unter', 'dem'] },  //(under the)
+  { word: 'vom', out: ['von', 'dem'] },  //(from the)
+  { word: 'vors', out: ['vor', 'das'] },  //(in front of the)
+  { word: 'vorm', out: ['vor', 'dem'] },  //(in front of the / from)
+  { word: 'zum', out: ['zu', 'dem'] },  //(to the / to)
+  { word: 'zur', out: ['zu', 'der'] },  //(to the / to)
+
+  // { word: "n", out: ['das', 'ans'] },
+  // { word: "an", out: ['dem', 'am'] },
+  // { word: "auf", out: ['das', 'aufs'] },
+  // { word: "bei", out: ['dem', 'beim'] },
+  // { word: "durch", out: ['das', 'durchs'] },
+  // { word: "für", out: ['das', 'fürs'] },
+  // { word: "in", out: ['das', 'ins'] },
+  // { word: "in", out: ['dem', 'im'] },
+  // { word: "über", out: ['das', 'übers'] },
+  // { word: "um", out: ['das', 'ums'] },
+  // { word: "unter", out: ['das', 'unters'] },
+  // { word: "von", out: ['dem', 'vom'] },
+  // { word: "vor", out: ['das', 'vors'] },
+  // { word: "vor", out: ['dem', 'vorm'] },
+  // { word: "zu", out: ['dem', 'zum'] },
+  // { word: "zu", out: ['der', 'zur'] },
+]
+
+

package/src/01-one/tokenizer/is-sentence.js

@@ -0,0 +1,37 @@
+const isAcronym = /[ .][A-Z]\.? *$/i
+const hasEllipse = /(?:\u2026|\.{2,}) *$/
+const hasLetter = /\p{L}/u
+const isOrdinal = /[0-9]\. *$/
+
+/** does this look like a sentence? */
+const isSentence = function (str, abbrevs) {
+  // must have a letter
+  if (hasLetter.test(str) === false) {
+    return false
+  }
+  // check for 'F.B.I.'
+  if (isAcronym.test(str) === true) {
+    return false
+  }
+  // german ordinals like '4.'
+  if (isOrdinal.test(str) === true) {
+    return false
+  }
+  //check for '...'
+  if (hasEllipse.test(str) === true) {
+    return false
+  }
+  let txt = str.replace(/[.!?\u203D\u2E18\u203C\u2047-\u2049] *$/, '')
+  let words = txt.split(' ')
+  let lastWord = words[words.length - 1].toLowerCase()
+  // check for 'Mr.'
+  if (abbrevs.hasOwnProperty(lastWord) === true) {
+    return false
+  }
+  // //check for jeopardy!
+  // if (blacklist.hasOwnProperty(lastWord)) {
+  //   return false
+  // }
+  return true
+}
+export default isSentence

package/src/01-one/tokenizer/plugin.js

@@ -0,0 +1,11 @@
+import unicode from './unicode.js'
+import contractions from './contractions.js'
+import isSentence from './is-sentence.js'
+
+export default {
+  mutate: (world) => {
+    world.model.one.unicode = unicode
+    world.model.one.contractions = contractions
+    world.methods.one.tokenize.isSentence = isSentence
+  },
+}

package/src/01-one/tokenizer/unicode.js

@@ -0,0 +1,55 @@
+//a hugely-ignorant, and widely subjective transliteration of latin, cryllic, greek unicode characters to english ascii.
+//approximate visual (not semantic or phonetic) relationship between unicode and ascii characters
+//http://en.wikipedia.org/wiki/List_of_Unicode_characters
+//https://docs.google.com/spreadsheet/ccc?key=0Ah46z755j7cVdFRDM1A2YVpwa1ZYWlpJM2pQZ003M0E
+
+
+// geman accented letters
+// ä  | Ä
+// ö  | Ö
+// ü  | Ü
+// ß
+
+let compact = {
+  '!': '¡',
+  '?': '¿Ɂ',
+  '"': '“”"❝❞',
+  "'": '‘‛❛❜’',
+  '-': '—–',
+  a: 'ªÁÃÅáãåĀāĂ㥹ǍǎǞǟǠǡǺǻȀȁȂȃȦȧȺΆΑΔΛάαλАаѦѧӐӑӒӓƛæ',
+  b: 'þƀƁƂƃƄƅɃϦБВЪЬвъьѢѣҌҍ',
+  c: '¢©ĆćĈĉĊċČčƆƇƈȻȼͻͼϲϹϽϾСсєҀҁҪҫ',
+  d: 'ÐĎďĐđƉƊȡƋƌ',
+  e: 'ĒēĔĕĖėĘęĚěƐȄȅȆȇȨȩɆɇΈΕΞΣέεξϵЀЁЕеѐёҼҽҾҿӖӗ',
+  f: 'ƑƒϜϝӺӻҒғſ',
+  g: 'ĜĝĞğĠġĢģƓǤǥǦǧǴǵ',
+  h: 'ĤĥĦħƕǶȞȟΉΗЂЊЋНнђћҢңҤҥҺһӉӊ',
+  I: 'Í',
+  i: 'íĨĩĪīĬĭĮįİıƖƗȈȉȊȋΊΐΪίιϊІЇії',
+  j: 'ĴĵǰȷɈɉϳЈј',
+  k: 'ĶķĸƘƙǨǩΚκЌЖКжкќҚқҜҝҞҟҠҡ',
+  l: 'ĹĺĻļĽľĿŀŁłƚƪǀǏǐȴȽΙӀӏ',
+  m: 'ΜϺϻМмӍӎ',
+  n: 'ÑñŃńŅņŇňʼnŊŋƝƞǸǹȠȵΝΠήηϞЍИЙЛПийлпѝҊҋӅӆӢӣӤӥπ',
+  o: 'ÓÕØðóõøŌōŎŏŐőƟƠơǑǒǪǫǬǭǾǿȌȍȎȏȪȫȬȭȮȯȰȱΌΘΟθοσόϕϘϙϬϴОФоѲѳӦӧӨөӪӫ',
+  p: 'ƤΡρϷϸϼРрҎҏÞ',
+  q: 'Ɋɋ',
+  r: 'ŔŕŖŗŘřƦȐȑȒȓɌɍЃГЯгяѓҐґ',
+  s: 'ŚśŜŝŞşŠšƧƨȘșȿЅѕ',
+  t: 'ŢţŤťŦŧƫƬƭƮȚțȶȾΓΤτϮТт',
+  u: 'µÚúŨũŪūŬŭŮůŰűŲųƯưƱƲǓǔǕǖǗǘǙǚǛǜȔȕȖȗɄΰμυϋύ',
+  v: 'νѴѵѶѷ',
+  w: 'ŴŵƜωώϖϢϣШЩшщѡѿ',
+  x: '×ΧχϗϰХхҲҳӼӽӾӿ',
+  y: 'ÝýÿŶŷŸƳƴȲȳɎɏΎΥΫγψϒϓϔЎУучўѰѱҮүҰұӮӯӰӱӲӳ',
+  z: 'ŹźŻżŽžƵƶȤȥɀΖ',
+  ß: 'β' //greek beta to german Eszett surrogate
+}
+//decompress data into two hashes
+let unicode = {}
+Object.keys(compact).forEach(function (k) {
+  compact[k].split('').forEach(function (s) {
+    unicode[s] = k
+  })
+})
+export default unicode

package/src/02-two/postTagger/matches.js

@@ -0,0 +1,47 @@
+const separablePrefixes = '(ab|an|auf|aus|bei|ein|fest|fort|her|heraus|hervor|hin|hinzu|los|mit|nach|nieder|vor|voraus|weg|weiter|zu|zurecht|zurück|zusammen|zuteil)'
+
+const postTagger = function (doc) {
+  // eine as 1 or the
+  doc.match('eine #Value').tag('TextValue', 'eine-value')
+  // 6.30 Uhr
+  doc.match('#Value uhr').tag('Time', 'time-Uhr')
+  // a value before a currency is money - '12,5 Euro'
+  doc.match('[#Value] #Currency', 0).tag('Money', 'value-currency')
+
+  // prenominal possessive is a determiner - 'mein Bruder', 'seine kleine Schwester'
+  doc.match('[(mein|dein|sein|ihr|unser|euer)] #Adjective? #Noun', 0).tag('Determiner', 'possessive-det')
+  doc.match('[#Possessive] #Adjective? #Noun', 0).tag('Determiner', 'possessive-det')
+
+  // 'weiß' beside a pronoun is wissen, not white - 'das weiß ich'
+  doc.match('[weiß] #Pronoun', 0).tag('PresentTense', 'weiß-pronoun')
+  doc.match('#Pronoun [weiß]', 0).tag('PresentTense', 'pronoun-weiß')
+
+  // a verb between a determiner and a noun is an inflected adjective - 'das schnelle Auto'
+  doc.match('#Determiner [(#Verb && !#Infinitive && /e[nrsm]?$/)] (#Noun && !#Pronoun)', 0).tag('Adjective', 'det-adj-noun')
+
+  // a determiner with no noun after it is a pronoun - 'das ist gut', 'alle außer ihm'
+  doc.match('[#Determiner] (#Verb|#Preposition)', 0).tag('Pronoun', 'det-pronoun')
+
+  // modal verbs - the infinitive goes to the end of the clause:
+  // dürfen (may), können (can), mögen (like), müssen (must), sollen (should), wollen (want)
+  doc.match('[{dürfen}] * #Infinitive', 0).tag('Modal', 'dürfen-verb')
+  doc.match('[{können}] * #Infinitive', 0).tag('Modal', 'können-verb')
+  doc.match('[{mögen}] * #Infinitive', 0).tag('Modal', 'mögen-verb')
+  doc.match('[{müssen}] * #Infinitive', 0).tag('Modal', 'müssen-verb')
+  doc.match('[{sollen}] * #Infinitive', 0).tag('Modal', 'sollen-verb')
+  doc.match('[{wollen}] * #Infinitive', 0).tag('Modal', 'wollen-verb')
+
+  // auxiliary + participle at the end of the clause (Satzklammer):
+  // 'sie hat gestern ein Buch gelesen', 'er ist nach Hause gegangen'
+  doc.match('[{haben}] * #Participle', 0).tag('Auxiliary', 'haben-participle')
+  doc.match('[{haben}] * #Infinitive', 0).tag('Auxiliary', 'haben-infinitive')
+  doc.match('[{sein}] * #Participle', 0).ifNo('#Determiner').tag('Auxiliary', 'sein-participle')
+  // werden + infinitive is the future tense - 'wir werden nach Berlin fahren'
+  doc.match('[{werden}] * #Infinitive', 0).tag('Auxiliary', 'werden-future')
+  // werden + participle is the passive - 'das Buch wird gelesen'
+  doc.match('[{werden}] * #Participle', 0).tag('Auxiliary', 'werden-passive')
+
+  // a separable verb-prefix, at the end of the clause - 'ich stehe um sieben Uhr auf'
+  doc.match('#Verb * [#Preposition]$', 0).match(separablePrefixes).tag('Particle', 'separable-prefix')
+}
+export default postTagger

package/src/02-two/postTagger/plugin.js

@@ -0,0 +1,8 @@
+import postTagger from './matches.js'
+
+export default {
+  compute: {
+    postTagger
+  },
+  hooks: ['postTagger']
+}

package/src/02-two/preTagger/compute/1st-pass/regex.js

@@ -0,0 +1,43 @@
+const hasApostrophe = /['‘’‛‵′`´]/
+const hasPeriod = /\./
+const isNum = /^[0-9+\-,]+$/
+
+// normal regexes
+const doRegs = function (str, regs) {
+  for (let i = 0; i < regs.length; i += 1) {
+    if (regs[i][0].test(str) === true) {
+      return regs[i]
+    }
+  }
+  return null
+}
+
+const checkRegex = function (terms, i, world) {
+  let setTag = world.methods.one.setTag
+  let term = terms[i]
+  let { regexText, regexNormal, regexNumbers } = world.model.two
+  let normal = term.machine || term.normal
+  let text = term.text
+  // keep dangling apostrophe?
+  if (hasApostrophe.test(term.post) && !hasApostrophe.test(term.pre)) {
+    text += term.post.trim()
+  }
+  // keep period in number ordinals?
+  if (hasPeriod.test(term.post) && isNum.test(text)) {
+    setTag([term], ['Ordinal', 'NumericValue'], world, false, `1-regex-ordinal`)
+    term.confidence = 0.6
+    return true
+  }
+  let arr = doRegs(text, regexText) || doRegs(normal, regexNormal)
+  // hide a bunch of number regexes behind this one
+  if (!arr && /[0-9]/.test(normal)) {
+    arr = doRegs(normal, regexNumbers)
+  }
+  if (arr) {
+    setTag([term], arr[1], world, false, `1-regex- '${arr[2] || arr[0]}'`)
+    term.confidence = 0.6
+    return true
+  }
+  return null
+}
+export default checkRegex

package/src/02-two/preTagger/compute/1st-pass/titlecase.js

@@ -0,0 +1,21 @@
+const isTitleCase = function (str) {
+  return /^[A-ZÄÖÜ][a-z'\u00C0-\u00FF]/.test(str) || /^[A-ZÄÖÜ]$/.test(str)
+}
+
+// add a noun to any non-0 index titlecased word, with no existing tag
+const titleCaseNoun = function (terms, i, world) {
+  let setTag = world.methods.one.setTag
+  let term = terms[i]
+  // don't over-write any tags
+  // if (term.tags.size > 0) {
+  //   return
+  // }
+  // skip first-word, for now
+  if (i === 0) {
+    return
+  }
+  if (isTitleCase(term.text)) {
+    setTag([term], 'Noun', world, false, `1-titlecase`)
+  }
+}
+export default titleCaseNoun

package/src/02-two/preTagger/compute/1st-pass/year.js

@@ -0,0 +1,54 @@
+const min = 1400
+const max = 2100
+
+const dateWords = new Set(['pendant', 'dans', 'avant', 'apres', 'pour', 'en'])
+
+const seemsGood = function (term) {
+  if (!term) {
+    return false
+  }
+  if (dateWords.has(term.normal)) {
+    return true
+  }
+  if (term.tags.has('Date') || term.tags.has('Month') || term.tags.has('WeekDay')) {
+    return true
+  }
+  return false
+}
+
+const seemsOkay = function (term) {
+  if (!term) {
+    return false
+  }
+  if (term.tags.has('Ordinal')) {
+    return true
+  }
+  return false
+}
+
+// recognize '1993' as a year
+const tagYear = function (terms, i, world) {
+  let setTag = world.methods.one.setTag
+  const term = terms[i]
+  if (term.tags.has('NumericValue') && term.tags.has('Cardinal') && term.normal.length === 4) {
+    let num = Number(term.normal)
+    // number between 1400 and 2100
+    if (num && !isNaN(num)) {
+      if (num > min && num < max) {
+        if (seemsGood(terms[i - 1]) || seemsGood(terms[i + 1])) {
+          setTag([term], 'Year', world, false, '1-tagYear')
+          return true
+        }
+        // or is it really-close to a year?
+        if (num > 1950 && num < 2025) {
+          if (seemsOkay(terms[i - 1]) || seemsOkay(terms[i + 1])) {
+            setTag([term], 'Year', world, false, '1-tagYear-close')
+            return true
+          }
+        }
+      }
+    }
+  }
+  return null
+}
+export default tagYear

package/src/02-two/preTagger/compute/2nd-pass/acronym.js

@@ -0,0 +1,82 @@
+const oneLetterAcronym = /^[A-ZÄÖÜ]('s|,)?$/
+const isUpperCase = /^[A-Z-ÄÖÜ]+$/
+const periodAcronym = /([A-ZÄÖÜ]\.)+[A-ZÄÖÜ]?,?$/
+const noPeriodAcronym = /[A-ZÄÖÜ]{2,}('s|,)?$/
+const lowerCaseAcronym = /([a-zäöü]\.)+[a-zäöü]\.?$/
+
+
+
+const oneLetterWord = {
+  I: true,
+  A: true,
+}
+// just uppercase acronyms, no periods - 'UNOCHA'
+const isNoPeriodAcronym = function (term, model) {
+  let str = term.text
+  // ensure it's all upper-case
+  if (isUpperCase.test(str) === false) {
+    return false
+  }
+  // long capitalized words are not usually either
+  if (str.length > 5) {
+    return false
+  }
+  // 'I' is not a acronym
+  if (oneLetterWord.hasOwnProperty(str)) {
+    return false
+  }
+  // known-words, like 'PIZZA' is not an acronym.
+  if (model.one.lexicon.hasOwnProperty(term.normal)) {
+    return false
+  }
+  //like N.D.A
+  if (periodAcronym.test(str) === true) {
+    return true
+  }
+  //like c.e.o
+  if (lowerCaseAcronym.test(str) === true) {
+    return true
+  }
+  //like 'F.'
+  if (oneLetterAcronym.test(str) === true) {
+    return true
+  }
+  //like NDA
+  if (noPeriodAcronym.test(str) === true) {
+    return true
+  }
+  return false
+}
+
+const isAcronym = function (terms, i, world) {
+  let setTag = world.methods.one.setTag
+  let term = terms[i]
+  //these are not acronyms
+  if (term.tags.has('RomanNumeral') || term.tags.has('Acronym')) {
+    return null
+  }
+  //non-period ones are harder
+  if (isNoPeriodAcronym(term, world.model)) {
+    term.tags.clear()
+    setTag([term], ['Acronym', 'Noun'], world, false, '3-no-period-acronym')
+    return true
+  }
+  // one-letter acronyms
+  if (!oneLetterWord.hasOwnProperty(term.text) && oneLetterAcronym.test(term.text)) {
+    term.tags.clear()
+    setTag([term], ['Acronym', 'Noun'], world, false, '3-one-letter-acronym')
+    return true
+  }
+  //if it's a very-short organization?
+  if (term.tags.has('Organization') && term.text.length <= 3) {
+    setTag([term], 'Acronym', world, false, '3-org-acronym')
+    return true
+  }
+  // upper-case org, like UNESCO
+  if (term.tags.has('Organization') && isUpperCase.test(term.text) && term.text.length <= 6) {
+    setTag([term], 'Acronym', world, false, '3-titlecase-acronym')
+    return true
+  }
+  return null
+}
+export default isAcronym

package/src/02-two/preTagger/compute/2nd-pass/fallback.js

@@ -0,0 +1,27 @@
+const isTitleCase = function (str) {
+  return /^[A-ZÄÖÜ][a-z'\u00C0-\u00FF]/.test(str) || /^[A-ZÄÖÜ]$/.test(str)
+}
+
+const hasNoVerb = function (terms) {
+  return !terms.find(t => t.tags.has('Verb'))
+}
+
+const fallback = function (terms, i, world) {
+  let setTag = world.methods.one.setTag
+  let term = terms[i]
+  if (term.tags.size === 0) {
+
+    // is it first-word titlecase?
+    if (i === 0 && isTitleCase(term.text)) {
+      setTag([term], 'Noun', world, false, `1-titlecase`)// Noun still the safest bet?
+      return
+    }
+
+    let tag = 'Adjective'
+    if (terms.length > 10 && hasNoVerb(terms)) {
+      tag = 'Verb'
+    }
+    setTag([term], tag, world, false, '2-fallback')
+  }
+}
+export default fallback