Initial tests for False Positives

lib/ofac/parsing.js

@@ -30,6 +30,10 @@ const partNames = new Map([
   [NICKNAME, 'nickname']
 ])
 
+const filteredWords = [
+  'al'
+]
+
 // group-id to type-id
 
 function processMasterNamePartGroup (groupNode) {
@@ -68,7 +72,17 @@ const processAlias = _.curry((groupTypes, aliasNode) => {
   const parts = _.map(getNamePart, namePartNodes)
 
   const fullName = nameUtils.makeFullName(parts)
-  const words = nameUtils.makeWords(fullName)
+  const words = _.flow(
+    nameUtils.makeWords,
+    _.reject(_.flow(
+      _.get('value'),
+      word => filteredWords.includes(word)
+    ))
+  )(fullName)
+
+  // if (words.length < 2) {
+  //   console.log(JSON.stringify(words))
+  // }
 
   return {id, parts, fullName, words}
 })

tests/ofac/matching.js

@@ -1,5 +1,7 @@
 const assert = require('assert')
 const ofac = require('../../lib/ofac')
+const fs = require('fs')
+const path = require('path')
 const _ = require('lodash/fp')
 
 let structs
@@ -206,5 +208,37 @@ describe('OFAC', function () {
       }
     })
 
+    it('should not match against common names', function () {
+      this.timeout(0)
+
+      const getNamesFromFile = _.flow(
+        name => path.resolve(__dirname, name),
+        file => fs.readFileSync(file, 'utf-8'),
+        _.split('\n'),
+        _.map( _.flow(
+          _.split(' '),
+          _.first
+        ))
+      )
+
+      const lastNames = getNamesFromFile('dist.all.last.txt')
+      const firstNamesMale = getNamesFromFile('dist.male.first.txt')
+      const firstNamesFemale = getNamesFromFile('dist.female.first.txt')
+
+      for (const lastName of lastNames.slice(0, 100)) {
+        for (firstName of firstNamesMale.slice(0, 100)) {
+          const matches = ofac.match({firstName, lastName}, null, 0.8)
+          console.log({firstName, lastName})
+          assert.ok(_.isEmpty(matches))
+        }
+
+        for (firstName of firstNamesFemale.slice(0, 100)) {
+          const matches = ofac.match({firstName, lastName}, null, 0.8)
+          console.log({firstName, lastName})
+          assert.ok(_.isEmpty(matches))
+        }
+      }
+    })
+
   })
 })