{ "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} } ], "sufficientConditions" : [ { "name" : "exactMatch", "fields" : [ "gridid" ] } ], "necessaryConditions" : [ { "name" : "DomainExactMatch", "fields" : [ "websiteurl" ] }, { "name" : "exactMatch", "fields" : [ "country" ] } ], "decisionTree" : { "start": {"fields": [{"field":"legalname", "comparator":"jaroWinklerNormalizedName", "weight":0.9, "ignoreMissing":"false", "params":{"windowSize" : 4, "threshold" : 0.7}}, {"field":"legalshortname", "comparator":"jaroWinklerNormalizedName", "weight":0.1, "ignoreMissing":"true", "params":{}}], "threshold": 0.9, "aggregation": "WEIGHTED_MEAN", "positive":"MATCH", "negative":"NO_MATCH", "undefined":"NO_MATCH", "ignoreMissing": "true"} }, "model" : [ { "name" : "country", "type" : "String", "path" : "organization/metadata/country/classid", "ignoreMissing" : "true" }, { "name" : "legalshortname", "type" : "String", "path" : "organization/metadata/legalshortname/value", "ignoreMissing" : "true" }, { "name" : "legalname", "type" : "String", "path" : "organization/metadata/legalname/value", "params" : {"windowSize" : 4, "threshold" : 0.7}, "ignoreMissing" : "false" }, { "name" : "websiteurl", "type" : "URL", "path" : "organization/metadata/websiteurl/value", "params" : { "host" : 0.5, "path" : 0.5 }, "ignoreMissing" : "true" }, { "name" : "gridid", "type" : "String", "path" : "pid[qualifier#classid = {grid}]/value", "ignoreMissing" : "true" } ], "blacklists" : { "legalname" : [] } } }