38 lines
2.5 KiB
Plaintext
38 lines
2.5 KiB
Plaintext
{
|
|
"wf" : {
|
|
"threshold" : "0.9",
|
|
"dedupRun" : "001",
|
|
"entityType" : "organization",
|
|
"orderField" : "legalname",
|
|
"queueMaxSize" : "2000",
|
|
"groupMaxSize" : "50",
|
|
"slidingWindowSize" : "200",
|
|
"rootBuilder" : [ "organization", "projectOrganization_participation_isParticipant", "datasourceOrganization_provision_isProvidedBy" ],
|
|
"includeChildren" : "true"
|
|
},
|
|
"pace" : {
|
|
"clustering" : [
|
|
{ "name" : "sortedngrampairs", "fields" : [ "legalname" ], "params" : { "max" : 2, "ngramLen" : "3"} },
|
|
{ "name" : "suffixprefix", "fields" : [ "legalname" ], "params" : { "max" : 1, "len" : "3" } },
|
|
{ "name" : "urlclustering", "fields" : [ "websiteurl" ], "params" : { } },
|
|
{ "name" : "keywordsclustering", "fields" : [ "legalname" ], "params" : { "max": 2, "windowSize": 4} }
|
|
],
|
|
"decisionTree" : {
|
|
"start": {"fields": [{"field":"gridid", "comparator":"exactMatch", "weight":1.0, "countIfUndefined":"true", "params": {}}], "threshold":1.0, "aggregation": "MAX", "positive":"MATCH", "negative":"layer2", "undefined":"layer2", "ignoreUndefined": "true"},
|
|
"layer2": {"fields": [{"field":"websiteurl", "comparator":"domainExactMatch", "weight":1.0, "countIfUndefined":"true", "params" : {}}, {"field":"country", "comparator":"exactMatch", "weight":1.0, "countIfUndefined":"false", "params": {}}], "threshold":1.0, "aggregation": "MIN", "positive":"layer3", "negative":"NO_MATCH", "undefined":"layer3", "ignoreUndefined": "false"},
|
|
"layer3": {"fields": [{"field":"legalname", "comparator":"jaroWinklerNormalizedName", "weight":0.9, "countIfUndefined":"false", "params":{"windowSize" : 4, "threshold" : 0.7}}, {"field":"legalshortname", "comparator":"jaroWinklerNormalizedName", "weight":0.1, "countIfUndefined":"true", "params":{}}], "threshold": 0.9, "aggregation": "W_MEAN", "positive":"MATCH", "negative":"NO_MATCH", "undefined":"NO_MATCH", "ignoreUndefined": "true"}
|
|
},
|
|
"model" : [
|
|
{ "name" : "country", "type" : "String", "path" : "organization/metadata/country/classid"},
|
|
{ "name" : "legalshortname", "type" : "String", "path" : "organization/metadata/legalshortname/value"},
|
|
{ "name" : "legalname", "type" : "String", "path" : "organization/metadata/legalname/value" },
|
|
{ "name" : "websiteurl", "type" : "URL", "path" : "organization/metadata/websiteurl/value" },
|
|
{ "name" : "gridid", "type" : "String", "path" : "pid[qualifier#classid = {grid}]/value"}
|
|
],
|
|
"blacklists" : {
|
|
"legalname" : []
|
|
},
|
|
"synonyms": {
|
|
}
|
|
}
|
|
} |