Merge pull request #30 from duckduckgo/dharb/july-entity-update

July entity update
duckduckgo · Jul 28, 2020 · 23be65e · 23be65e
2 parents 364c337 + 5c06f92
commit 23be65e
Show file tree

Hide file tree

Showing 49 changed files with 8,232 additions and 2,277 deletions.
diff --git a/build-data/generated/domain_map.json b/build-data/generated/domain_map.json
diff --git a/build-data/generated/entity_map.json b/build-data/generated/entity_map.json
diff --git a/entities/Airbnb, Inc..json b/entities/Airbnb, Inc..json
@@ -3,6 +3,7 @@
     "displayName": "Airbnb",
     "properties": [
         "abnb.me",
+        "air.ml",
         "air.tl",
         "airbnb.ae",
         "airbnb.al",
@@ -60,6 +61,7 @@
         "airbnb.fr",
         "airbnb.gr",
         "airbnb.gy",
+        "airbnb.hosting",
         "airbnb.hu",
         "airbnb.io",
         "airbnb.is",

diff --git a/entities/AppNexus, Inc..json b/entities/AppNexus, Inc..json
diff --git a/entities/Charter Communications Operating, LLC.json b/entities/Charter Communications Operating, LLC.json
diff --git a/entities/Charter Communications, Inc..json b/entities/Charter Communications, Inc..json
@@ -0,0 +1,23 @@
+{
+    "name": "Charter Communications, Inc.",
+    "displayName": "Charter Communications",
+    "properties": [
+        "baynews9.com",
+        "brighthouse.com",
+        "charter.com",
+        "charterbusiness.com",
+        "charterverify.com",
+        "mynews13.com",
+        "ny1.com",
+        "roadrunner.com",
+        "spectrum.com",
+        "spectrum.net",
+        "spectrumbiz.com",
+        "spectrumenterprise.com",
+        "spectrumlocalnews.com",
+        "spectrummobile.com",
+        "timewarnercable.com",
+        "twc.com",
+        "twcc.com"
+    ]
+}
diff --git a/entities/Department of Premier and Cabinet.json b/entities/Department of Premier and Cabinet.json
diff --git a/entities/Department of the Premier and Cabinet.json b/entities/Department of the Premier and Cabinet.json
@@ -2,6 +2,7 @@
     "name": "Department of the Premier and Cabinet",
     "displayName": "Department of the Premier and Cabinet",
     "properties": [
-        "www.sa.gov.au"
+        "sa.gov.au",
+        "vic.gov.au"
     ]
 }
diff --git a/entities/Facebook, Inc..json b/entities/Facebook, Inc..json
@@ -8,8 +8,10 @@
         "atlassbx.com",
         "atlassolutions.com",
         "cdninstagram.com",
+        "crowdtangle.com",
         "facebook.com",
         "facebook.net",
+        "facebookmail.com",
         "fb.com",
         "fb.gg",
         "fb.me",
@@ -31,12 +33,18 @@
         "oculuscdn.com",
         "oculusrift.com",
         "oculusvr.com",
+        "onavo.com",
+        "onavo.net",
+        "onavo.org",
+        "onavoinsights.com",
         "powersunitedvr.com",
         "reactjs.org",
         "thefind.com",
+        "vircado.com",
         "wa.me",
         "whatsapp.com",
-        "whatsapp.net"
+        "whatsapp.net",
+        "wit.ai"
     ],
     "prevalence": {
         "tracking": 0.353,

diff --git a/entities/Flowplayer AB.json b/entities/Flowplayer AB.json
@@ -2,7 +2,8 @@
     "name": "Flowplayer AB",
     "displayName": "Flowplayer",
     "properties": [
-        "flowplayer.com"
+        "flowplayer.com",
+        "flowplayer.org"
     ],
     "prevalence": {
         "tracking": 0.000151,

diff --git a/entities/Flowzymes Ky.json b/entities/Flowzymes Ky.json
diff --git a/entities/GCIS.json b/entities/GCIS.json
@@ -2,6 +2,6 @@
     "name": "GCIS",
     "displayName": "GCIS",
     "properties": [
-        "www.gov.za"
+        "gov.za"
     ]
 }
diff --git a/entities/GitHub, Inc..json b/entities/GitHub, Inc..json
@@ -8,7 +8,9 @@
         "github.io",
         "githubapp.com",
         "githubassets.com",
-        "githubusercontent.com"
+        "githubusercontent.com",
+        "npmjs.com",
+        "npmjs.org"
     ],
     "prevalence": {
         "tracking": 0.000958,

diff --git a/entities/Gmina Wroclaw.json b/entities/Gmina Wroclaw.json
@@ -2,6 +2,6 @@
     "name": "Gmina Wroclaw",
     "displayName": "Gmina Wroclaw",
     "properties": [
-        "www.wroclaw.pl"
+        "wroclaw.pl"
     ]
 }
diff --git a/entities/Go Daddy Operating Company, LLC.json b/entities/Go Daddy Operating Company, LLC.json
diff --git a/entities/GoDaddy Inc..json b/entities/GoDaddy Inc..json
@@ -0,0 +1,12 @@
+{
+    "name": "GoDaddy Inc.",
+    "displayName": "GoDaddy",
+    "properties": [
+        "godaddy.com",
+        "madmimi.com",
+        "myftpupload.com",
+        "secureserver.net",
+        "starfieldtech.com",
+        "sucuri.net"
+    ]
+}
diff --git a/entities/GoDaddy.json b/entities/GoDaddy.json
diff --git a/entities/Government Digital Service.json b/entities/Government Digital Service.json
@@ -3,7 +3,7 @@
     "displayName": "Government Digital Service",
     "properties": [
         "charitycommission.gov.uk",
-        "service.gov.uk",
-        "www.gov.uk"
+        "gov.uk",
+        "service.gov.uk"
     ]
 }
diff --git a/entities/Grupa Interia.pl Sp. z o.o. sp. k..json b/entities/Grupa Interia.pl Sp. z o.o. sp. k..json
@@ -2,12 +2,23 @@
     "name": "Grupa Interia.pl Sp. z o.o. sp. k.",
     "displayName": "Interia",
     "properties": [
+        "adretail.pl",
         "adsearch.pl",
         "bryk.pl",
+        "czateria.pl",
+        "deccoria.pl",
         "ding.pl",
+        "esporter.pl",
         "interia.pl",
         "iplsc.com",
-        "smaker.pl"
+        "maxmodels.pl",
+        "opracowania.pl",
+        "pomponik.pl",
+        "promocyjni.pl",
+        "smaker.pl",
+        "styl.pl",
+        "swiatseriali.pl",
+        "tipy.pl"
     ],
     "prevalence": {
         "tracking": 0.0000684,

diff --git a/entities/Meredith Corporation.json b/entities/Meredith Corporation.json
@@ -20,20 +20,36 @@
         "allrecipes.ru",
         "ar-cdn.com",
         "bhg.com",
+        "coastalliving.com",
+        "cookinglight.com",
         "divinecaroline.com",
         "eatingwell.com",
         "fitnessmagazine.com",
+        "foodandwine.com",
+        "fortune.com",
+        "hellogiggles.com",
+        "instyle.com",
         "media-allrecipes.com",
         "meredith.com",
         "midwestliving.com",
         "more.com",
+        "myrecipes.com",
         "mywedding.com",
         "parenting.com",
         "parents.com",
+        "people.com",
         "peopleenespanol.com",
+        "realsimple.com",
         "selectablemedia.com",
         "shape.com",
+        "si.com",
+        "southernliving.com",
+        "time.com",
+        "timeinc.com",
+        "timeinc.net",
+        "timeincapp.com",
         "traditionalhome.com",
+        "travelandleisure.com",
         "woodmagazine.com"
     ],
     "prevalence": {

diff --git a/entities/Office of Superintendent of Public Instruction.json b/entities/Office of Superintendent of Public Instruction.json
@@ -2,6 +2,6 @@
     "name": "Office of Superintendent of Public Instruction",
     "displayName": "Office of Superintendent of Public Instruction",
     "properties": [
-        "www.k12.wa.us"
+        "k12.wa.us"
     ]
 }
diff --git a/entities/Oracle Corporation.json b/entities/Oracle Corporation.json
@@ -37,6 +37,7 @@
         "oracleoutsourcing.com",
         "q-go.net",
         "responsys.net",
+        "rightnow.com",
         "rightnowtech.com",
         "rnengage.com",
         "univide.com",

diff --git a/entities/Pinduoduo Inc..json b/entities/Pinduoduo Inc..json
@@ -2,7 +2,9 @@
     "name": "Pinduoduo Inc.",
     "displayName": "Pinduoduo",
     "properties": [
+        "4pn.cn",
         "pinduoduo.com",
+        "y4n.cn",
         "yangkeduo.com"
     ]
 }
diff --git a/entities/Secretaria de Gobernacion.json b/entities/Secretaria de Gobernacion.json
@@ -2,6 +2,7 @@
     "name": "Secretaria de Gobernacion",
     "displayName": "Secretaria de Gobernacion",
     "properties": [
+        "curp.gob.mx",
         "dof.gob.mx"
     ]
 }
diff --git a/entities/Secretaria de Gobernaci�n.json b/entities/Secretaria de Gobernaci�n.json