rework crawler; add atom link extraction

[phinde.git] / data / elasticsearch-mapping.json

diff --git a/data/elasticsearch-mapping.json b/data/elasticsearch-mapping.json
index ec9bcc9665611af4981649f0255e12c25590e43a..d1e83ec4cf4eb4a1bb6e695b834a901e9cb2927e 100644 (file)
--- a/data/elasticsearch-mapping.json
+++ b/data/elasticsearch-mapping.json
@@ -20,11 +20,18 @@
                 },
                 "url": {
                     "type": "string",
-                    "index": "not_analyzed"
+                    "index": "not_analyzed",
+                    "boost": 1.5
+                },
+                "schemalessUrl": {
+                    "type": "string",
+                    "index": "not_analyzed",
+                    "boost": 1.5
                 },
                 "domain": {
                     "type": "string",
-                    "index": "not_analyzed"
+                    "index": "not_analyzed",
+                    "boost": 1.8
                 },
                 "source": {
                     "type": "string",
@@ -33,6 +40,42 @@
                 "language": {
                     "type": "string",
                     "index": "not_analyzed"
+                },
+                "title": {
+                    "type": "string",
+                    "boost": 2
+                },
+                "h1": {
+                    "type": "string",
+                    "boost": 1.8
+                },
+                "h2": {
+                    "type": "string",
+                    "boost": 1.7
+                },
+                "h3": {
+                    "type": "string",
+                    "boost": 1.6
+                },
+                "h4": {
+                    "type": "string",
+                    "boost": 1.5
+                },
+                "h5": {
+                    "type": "string",
+                    "boost": 1.4
+                },
+                "h6": {
+                    "type": "string",
+                    "boost": 1.3
+                },
+                "text": {
+                    "type": "string",
+                    "boost": 1.0
+                },
+                "tags": {
+                    "type": "string",
+                    "boost": 1.5
                 }
             }
         }