feat(venue_popularity): foundations for popularity scoring module

stream/importPipeline.js

@@ -14,6 +14,7 @@ streams.adminLookup = require('pelias-wof-admin-lookup').create;
 streams.addressExtractor = require('./address_extractor');
 streams.categoryMapper = require('./category_mapper');
 streams.addendumMapper = require('./addendum_mapper');
+streams.popularityMapper = require('./popularity_mapper');
 streams.dbMapper = require('pelias-model').createDocumentMapperStream;
 streams.elasticsearch = require('pelias-dbclient');
 
@@ -26,6 +27,7 @@ streams.import = function(){
     .pipe( streams.blacklistStream() )
     .pipe( streams.categoryMapper( categoryDefaults ) )
     .pipe( streams.addendumMapper() )
+    .pipe( streams.popularityMapper() )
     .pipe( streams.adminLookup() )
     .pipe( streams.dbMapper() )
     .pipe( streams.elasticsearch({name: 'openstreetmap'}) );

stream/popularity_mapper.js

@@ -0,0 +1,156 @@
+/**
+  The popularity mapper is responsible for generating a 'popularity'
+  value by inspecting OSM tags.
+
+  Feel free to make changes to this mapping file!
+**/
+
+const through = require('through2');
+const peliasLogger = require('pelias-logger').get('openstreetmap');
+
+const config = {
+  // https://taginfo.openstreetmap.org/keys/importance
+  importance: {
+    international: { _score: 20000 },
+    national: { _score: 5000 }
+  },
+  // https://taginfo.openstreetmap.org/keys/wikipedia
+  wikipedia: { _score: 2000 },
+  // https://taginfo.openstreetmap.org/keys/wikidata
+  wikidata: { _score: 2000 },
+
+  // misc properties found on major landmarks &
+  // public buildings.
+  architect: { _score: 5000 },
+  heritage: { _score: 5000 },
+  historic: { _score: 5000 },
+  building: {
+    colour: { _score: 2000 },
+    material: { _score: 2000 },
+    supermarket: { _score: 2000 },
+    civic: { _score: 2000 },
+    government: { _score: 2000 },
+    hospital: { _score: 2000 },
+    train_station: { _score: 5000 },
+    transportation: { _score: 5000 },
+    university: { _score: 2000 },
+    public: { _score: 2000 },
+    sports_hall: { _score: 2000 },
+  },
+  height: { _score: 2000 },
+  start_date: { _score: 2000 },
+
+  // misc properties found on tourist attractions
+  tourism: {
+    visitors: { _score: 2000 },
+    aquarium: { _score: 2000 },
+    attraction: { _score: 1000 },
+    museum: { _score: 2000 },
+    theme_park: { _score: 2000 },
+    zoo: { _score: 2000 },
+    _score: 1000
+  },
+  amenity: {
+    college: { _score: 2000 },
+    library: { _score: 2000 },
+    school: { _score: 1000 },
+    university: { _score: 2000 },
+    bank: { _score: 1000 },
+    clinic: { _score: 1000 },
+    dentist: { _score: 1000 },
+    doctors: { _score: 1000 },
+    hospital: { _score: 2000 },
+    nursing_home: { _score: 1000 },
+    pharmacy: { _score: 1000 },
+    social_facility: { _score: 1000 },
+    veterinary: { _score: 1000 },
+    community_centre: { _score: 1000 },
+    music_venue: { _score: 1000 },
+    nightclub: { _score: 1000 },
+    planetarium: { _score: 1000 },
+    social_centre: { _score: 1000 },
+    theatre: { _score: 1000 },
+    courthouse: { _score: 1000 },
+    coworking_space: { _score: 1000 },
+    dojo: { _score: 1000 },
+    embassy: { _score: 1000 },
+    fire_station: { _score: 1000 },
+    marketplace: { _score: 1000 },
+    place_of_worship: { _score: 1000 },
+    police: { _score: 1000 },
+    post_office: { _score: 1000 },
+    prison: { _score: 1000 },
+    public_bath: { _score: 1000 },
+    townhall: { _score: 1000 }
+  },
+  museum: { _score: 2000 },
+  museum_type: { _score: 2000 },
+  opening_hours: { _score: 1000 },
+  operator: { _score: 1000 },
+  fee: { _score: 2000 },
+  website: { _score: 2000 },
+  contact: {
+    website: { _score: 2000 },
+    email: { _score: 1000 },
+    phone: { _score: 1000 },
+    fax: { _score: 1000 },
+    foursquare: { _score: 2000 },
+    facebook: { _score: 2000 },
+    linkedin: { _score: 2000 },
+    instagram: { _score: 2000 },
+    skype: { _score: 2000 },
+    flickr: { _score: 2000 },
+    youtube: { _score: 2000 },
+  }
+};
+
+module.exports = function(){
+
+  return through.obj(( doc, enc, next ) => {
+
+    try {
+
+      // skip records with no tags
+      let tags = doc.getMeta('tags');
+      if( !tags ){
+        return next( null, doc );
+      }
+
+      // default popularity
+      let popularity = doc.getPopularity() || 0;
+
+      // apply scores from config
+      for( let tag in config ){
+        if( tags.hasOwnProperty( tag ) ){
+          // global score for the tag
+          if( config[tag]._score ){
+            popularity += config[tag]._score;
+          }
+          // individual scores for specific values
+          for( let value in config[tag] ){
+            if( value === '_score' ){ continue; }
+            if( !config[tag][value]._score ){ continue; }
+            if( tags[tag] === value.trim() ){
+              popularity += config[tag][value]._score;
+            }
+          }
+        }
+      }
+
+      // set document popularity when it is greater than zero
+      if( !!popularity ){
+        doc.setPopularity( popularity );
+      }
+    }
+
+    catch( e ){
+      peliasLogger.error( 'popularity_mapper error' );
+      peliasLogger.error( e.stack );
+      peliasLogger.error( JSON.stringify( doc, null, 2 ) );
+    }
+
+    return next( null, doc );
+
+  });
+
+};

test/end-to-end.js

@@ -52,6 +52,7 @@ streams.pbfParser()
   .pipe( streams.addressExtractor() )
   .pipe( streams.categoryMapper( streams.config.categoryDefaults ) )
   .pipe( streams.addendumMapper() )
+  .pipe( streams.popularityMapper() )
   .pipe( model.createDocumentMapperStream() )
   .pipe( sink.obj(function (doc) {
     results.push(doc);