Merge pull request #214 from pelias/avoid-transliterated-names

skip names we suspect were sourced from machine transliteration

prototype/wof.js

@@ -6,6 +6,7 @@ const util = require('util');
 const blacklist = require('pelias-blacklist-stream/loader')();
 const analysis = require('../lib/analysis');
 const language = dir('../config/language');
+const LOW_POPULATION_THRESHOLD = 2000;
 
 // list of languages / tags we favour in cases of deduplication
 const LANG_PREFS = ['eng','und'];
@@ -97,12 +98,25 @@ function insertWofRecord( wof, next ){
       }
     }
 
-    // add 'name:*'
-    for( var attr in wof ){
-      // https://github.com/whosonfirst/whosonfirst-names
-      // names: preferred|colloquial|variant|unknown
-      var match = attr.match(/^name:([a-z]{3})_x_(preferred|colloquial|variant)$/);
-      if( match ){
+    // note: skip all `name:*` fields when we suspect that they were sourced from
+    // machine transliteration via WikiData.
+    // see: https://github.com/whosonfirst-data/whosonfirst-data/issues/799
+    const hasDeadOrObscureLanguages = _.has(wof, 'name:vol_x_preferred');
+    const isLowOrUnknownPopulation = _.get(doc, 'population', 0) < LOW_POPULATION_THRESHOLD;
+    const isMegaCity = _.get(doc, 'wof:megacity', 0) === 1;
+    const isCapitalCity = !_.isEmpty(_.get(doc, 'wof:capital_of'));
+    const isLikelyTransliterated = (
+      hasDeadOrObscureLanguages && isLowOrUnknownPopulation && !isMegaCity && !isCapitalCity
+    );
+    if (!isLikelyTransliterated) {
+
+      // add 'name:*' fields
+      for( var attr in wof ){
+        // https://github.com/whosonfirst/whosonfirst-names
+        // names: preferred|colloquial|variant|unknown
+        const match = attr.match(/^name:([a-z]{3})_x_(preferred|colloquial|variant)$/);
+        if (!match) { continue; }
+
         // Fix for https://github.com/pelias/placeholder/pull/126
         // Transform iso codes 639-2/B to 639-2/T
         const lang = language.alternatives[match[1]] || match[1];
@@ -127,6 +141,7 @@ function insertWofRecord( wof, next ){
           doc.names[ lang ] = wof[ attr ];
         }
       }
+
     }
   }