dnet-dedup/dnet-dedup-test/src/main/resources/eu/dnetlib/pace/organization.test2.pace.conf

33 lines
1.8 KiB
Plaintext

{
"wf" : {
"threshold" : "0.98",
"dedupRun" : "001",
"entityType" : "organization",
"orderField" : "legalname",
"queueMaxSize" : "2000",
"groupMaxSize" : "10",
"slidingWindowSize" : "200",
"rootBuilder" : [ "organization", "projectOrganization_participation_isParticipant", "datasourceOrganization_provision_isProvidedBy" ],
"includeChildren" : "true"
},
"pace" : {
"clustering" : [
{ "name" : "ngrampairs", "fields" : [ "legalname" ], "params" : { "max" : 1, "ngramLen" : "3"} },
{ "name" : "suffixprefix", "fields" : [ "legalname" ], "params" : { "max" : 1, "len" : "3" } },
{ "name" : "urlclustering", "fields" : [ "websiteurl" ], "params" : { } }
],
"necessaryConditions" : [
{ "name" : "exactMatch", "fields" : [ "country" ] },
{ "name" : "DomainExactMatch", "fields" : [ "websiteurl" ] }
],
"decisionTree": {},
"model" : [
{ "name" : "legalname", "algo" : "Null", "type" : "String", "weight" : "0", "ignoreMissing" : "false", "path" : "organization/metadata/legalname/value" },
{ "name" : "country", "algo" : "Null", "type" : "String", "weight" : "0", "ignoreMissing" : "true", "path" : "organization/metadata/country/classid" },
{ "name" : "legalshortname", "algo" : "JaroWinkler", "type" : "String", "weight" : "0.3", "ignoreMissing" : "true", "path" : "organization/metadata/legalshortname/value" },
{ "name" : "legalname", "algo" : "JaroWinklerNormalizedName", "type" : "String", "weight" : "0.7", "ignoreMissing" : "false", "path" : "organization/metadata/legalname/value", "params" : { "windowSize" : 4 } },
{ "name" : "websiteurl", "algo" : "Null", "type" : "URL", "weight" : "0", "ignoreMissing" : "true", "path" : "organization/metadata/websiteurl/value", "params" : { "host" : 0.5, "path" : 0.5 } }
],
"blacklists" : { }
}
}