|
1
|
{
|
|
2
|
"wf" : {
|
|
3
|
"threshold" : "0.9",
|
|
4
|
"dedupRun" : "001",
|
|
5
|
"entityType" : "organization",
|
|
6
|
"orderField" : "legalname",
|
|
7
|
"queueMaxSize" : "2000",
|
|
8
|
"groupMaxSize" : "50",
|
|
9
|
"slidingWindowSize" : "200",
|
|
10
|
"rootBuilder" : [ "organization", "projectOrganization_participation_isParticipant", "datasourceOrganization_provision_isProvidedBy" ],
|
|
11
|
"includeChildren" : "true"
|
|
12
|
},
|
|
13
|
"pace" : {
|
|
14
|
"clustering" : [
|
|
15
|
{ "name" : "sortedngrampairs", "fields" : [ "legalname" ], "params" : { "max" : 2, "ngramLen" : "3"} },
|
|
16
|
{ "name" : "suffixprefix", "fields" : [ "legalname" ], "params" : { "max" : 1, "len" : "3" } },
|
|
17
|
{ "name" : "urlclustering", "fields" : [ "websiteurl" ], "params" : { } }
|
|
18
|
],
|
|
19
|
"strictConditions" : [
|
|
20
|
{ "name" : "exactMatch", "fields" : [ "gridid" ] }
|
|
21
|
],
|
|
22
|
"conditions" : [
|
|
23
|
{ "name" : "exactMatch", "fields" : [ "country" ] },
|
|
24
|
{ "name" : "DomainExactMatch", "fields" : [ "websiteurl" ] }
|
|
25
|
],
|
|
26
|
"model" : [
|
|
27
|
{ "name" : "country", "algo" : "Null", "type" : "String", "weight" : "0", "ignoreMissing" : "false", "path" : "organization/metadata/country/classid" },
|
|
28
|
{ "name" : "legalshortname", "algo" : "JaroWinklerNormalizedName", "type" : "String", "weight" : "0.1", "ignoreMissing" : "false", "path" : "organization/metadata/legalshortname/value" },
|
|
29
|
{ "name" : "legalname", "algo" : "JaroWinklerNormalizedName", "type" : "String", "weight" : "0.9", "ignoreMissing" : "false", "path" : "organization/metadata/legalname/value", "params" : {"windowSize" : 4, "threshold" : 0.5} },
|
|
30
|
{ "name" : "websiteurl", "algo" : "Null", "type" : "URL", "weight" : "0", "ignoreMissing" : "true", "path" : "organization/metadata/websiteurl/value", "params" : { "host" : 0.5, "path" : 0.5 } },
|
|
31
|
{ "name" : "gridid", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid[qualifier#classid = {grid}]/value" }
|
|
32
|
],
|
|
33
|
"blacklists" : { }
|
|
34
|
}
|
|
35
|
}
|