Parsing moved to downloading. Matching is being tweaked.

lib/ofac/matching.js

@@ -28,7 +28,8 @@ const isBornTooLongSince = _.curry((days, dateObject, individual) => {
 
 // algorithm
 
-function match (structs, candidate, threshold) {
+function match (structs, candidate, options) {
+  const {threshold, ratio = 0.1, debug, verboseFor} = options
   const {fullName, words, birthDate} = candidate
 
   // Accept aliases who's full name matches.
@@ -44,42 +45,57 @@ function match (structs, candidate, threshold) {
   )(aliases)
 
 
-  const aliasIds = []
-  const phoneticWeight = 0.17
+  const aliasIdCounts = new Map()
+  const phoneticWeight = ratio
   const stringWeight = 1 - phoneticWeight
 
   for (const word of words) {
     const getPhonetic = phonetic => structs.phoneticMap.get(phonetic)
     const phoneticMatches = new Set(_.flatMap(getPhonetic, word.phonetics))
 
+    const aliasIds = new Set()
+
     for (const wordEntry of structs.wordList) {
       const stringScore = stringSimilarity(word.value, wordEntry.value)
 
-      if (stringWeight * stringScore + phoneticWeight < threshold) continue
+      const verbose = _.includes(wordEntry.value, verboseFor)
+
+      if (!verbose && stringWeight * stringScore + phoneticWeight < threshold) continue
 
       for (const aliasId of wordEntry.aliasIds) {
-        const phoneticScore = phoneticMatches.has(aliasId) ? 1 : 0
-        const finalScore = stringWeight * stringScore + phoneticWeight * phoneticScore
+        const phoneticScore = phoneticMatches.has(aliasId) ? 1 : -1
+        // const finalScore = stringWeight * stringScore + phoneticWeight * phoneticScore
+        const finalScore = stringScore + phoneticWeight * phoneticScore
+
+        verbose && console.log(finalScore.toFixed(2), stringScore.toFixed(2), phoneticScore.toFixed(2), word.value, wordEntry.value)
 
         if (finalScore >= threshold) {
-          aliasIds.push(aliasId)
+          aliasIds.add(aliasId)
         }
       }
     }
+
+    verboseFor && console.log(aliasIds)
+
+    for (const aliasId of aliasIds.values()) {
+      const count = aliasIdCounts.get(aliasId) || 0
+      aliasIdCounts.set(aliasId, count + 1)
+    }
   }
 
-  const aliasIdsFromNamePart = _.flow(
-    _.countBy(_.identity),
-    _.toPairs,
-    _.reject(_.flow(
-      _.last,
-      _.gt(2)
-    )),
-    _.map(_.first)
-  )(aliasIds)
+  verboseFor && console.log(aliasIdCounts)
 
-  // debug_log(aliasIdsFromFullName)
-  // debug_log(aliasIdsFromNamePart)
+  const aliasIdsFromNamePart = []
+
+  for (const [aliasId, count] of aliasIdCounts) {
+    const {length} = structs.aliasesMap.get(aliasId).words
+    if (count >= _.min([2, words.length, length])) {
+      aliasIdsFromNamePart.push(aliasId)
+    }
+  }
+
+  debug && debug_log(aliasIdsFromFullName)
+  debug && debug_log(aliasIdsFromNamePart)
 
   // Get the full record for each matched id
   const getIndividual = aliasId => {