156 lines
4.6 KiB
JSON
156 lines
4.6 KiB
JSON
{
|
|
"wf" : {
|
|
"threshold" : "0.99",
|
|
"dedupRun" : "001",
|
|
"entityType" : "organization",
|
|
"orderField" : "legalname",
|
|
"queueMaxSize" : "2000",
|
|
"groupMaxSize" : "50",
|
|
"slidingWindowSize" : "200",
|
|
"idPath":"$.id",
|
|
"rootBuilder" : [ "organization", "projectOrganization_participation_isParticipant", "datasourceOrganization_provision_isProvidedBy" ],
|
|
"includeChildren" : "true",
|
|
"maxIterations": "20"
|
|
},
|
|
"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,
|
|
"countIfUndefined": "false",
|
|
"params": {}
|
|
}
|
|
],
|
|
"threshold": 1,
|
|
"aggregation": "AVG",
|
|
"positive": "MATCH",
|
|
"negative": "NO_MATCH",
|
|
"undefined": "layer2",
|
|
"ignoreUndefined": "false"
|
|
},
|
|
"layer2": {
|
|
"fields": [
|
|
{
|
|
"field": "websiteurl",
|
|
"comparator": "domainExactMatch",
|
|
"weight": 1,
|
|
"countIfUndefined": "false",
|
|
"params": {}
|
|
},
|
|
{
|
|
"field": "country",
|
|
"comparator": "exactMatch",
|
|
"weight": 1,
|
|
"countIfUndefined": "true",
|
|
"params": {}
|
|
},
|
|
{
|
|
"field": "legalname",
|
|
"comparator": "numbersMatch",
|
|
"weight": 1,
|
|
"countIfUndefined": "true",
|
|
"params": {}
|
|
},
|
|
{
|
|
"field": "legalname",
|
|
"comparator": "romansMatch",
|
|
"weight": 1,
|
|
"countIfUndefined": "true",
|
|
"params": {}
|
|
}
|
|
],
|
|
"threshold": 1,
|
|
"aggregation": "AND",
|
|
"positive": "layer3",
|
|
"negative": "NO_MATCH",
|
|
"undefined": "layer3",
|
|
"ignoreUndefined": "true"
|
|
},
|
|
"layer3": {
|
|
"fields": [
|
|
{
|
|
"field": "legalname",
|
|
"comparator": "cityMatch",
|
|
"weight": 1.0,
|
|
"countIfUndefined": "true",
|
|
"params": {
|
|
"windowSize": "4"
|
|
}
|
|
}
|
|
],
|
|
"threshold": 0.7,
|
|
"aggregation": "W_MEAN",
|
|
"positive": "layer4",
|
|
"negative": "NO_MATCH",
|
|
"undefined": "NO_MATCH",
|
|
"ignoreUndefined": "true"
|
|
},
|
|
"layer4": {
|
|
"fields": [
|
|
{
|
|
"field": "legalname",
|
|
"comparator": "keywordMatch",
|
|
"weight": 1.0,
|
|
"countIfUndefined": "true",
|
|
"params": {
|
|
"windowSize": "4"
|
|
}
|
|
}
|
|
],
|
|
"threshold": 0.9,
|
|
"aggregation": "AVG",
|
|
"positive": "layer5",
|
|
"negative": "NO_MATCH",
|
|
"undefined": "layer5",
|
|
"ignoreUndefined": "true"
|
|
},
|
|
"layer5": {
|
|
"fields": [
|
|
{
|
|
"field": "legalname",
|
|
"comparator": "jaroWinklerNormalizedName",
|
|
"weight": 0.9,
|
|
"countIfUndefined": "true",
|
|
"params": {
|
|
"windowSize": "4"
|
|
}
|
|
},
|
|
{
|
|
"field": "legalshortname",
|
|
"comparator": "jaroWinklerNormalizedName",
|
|
"weight": 0.1,
|
|
"countIfUndefined": "false",
|
|
"params": {}
|
|
}
|
|
],
|
|
"threshold": 0.99,
|
|
"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.ac')].value"},
|
|
{ "name" : "originalId", "type" : "String", "path" : "$.id" }
|
|
],
|
|
"blacklists" : {
|
|
"legalname" : []
|
|
},
|
|
"synonyms": {}
|
|
}
|
|
} |