52 lines
2.9 KiB
Plaintext
52 lines
2.9 KiB
Plaintext
{
|
|
"wf" : {
|
|
"threshold" : "0.99",
|
|
"dedupRun" : "001",
|
|
"entityType" : "result",
|
|
"orderField" : "title",
|
|
"queueMaxSize" : "2000",
|
|
"groupMaxSize" : "10",
|
|
"slidingWindowSize" : "200",
|
|
"rootBuilder" : [ "result" ],
|
|
"includeChildren" : "true"
|
|
},
|
|
"pace" : {
|
|
"clustering" : [
|
|
{ "name" : "acronyms", "fields" : [ "title" ], "params" : { "max" : "1", "minLen" : "2", "maxLen" : "4"} },
|
|
{ "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
|
|
{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } }
|
|
],
|
|
"decisionTree": {
|
|
"start": {"fields": [{"field":"pid", "comparator":"exactMatch", "weight":1.0, "countIfUndefined":"true", "params": {}}], "threshold":1.0, "aggregation": "MAX", "positive":"MATCH", "negative":"layer2", "undefined":"layer2", "ignoreUndefined": "true"},
|
|
"layer2": {"fields": [{"field":"dateofacceptance", "comparator":"yearMatch", "weight":1.0, "countIfUndefined":"true", "params" : {}}, {"field":"title", "comparator":"titleVersionMatch", "weight":1.0, "countIfUndefined":"false", "params": {}}, {"field":"authors", "comparator":"sizeMatch", "weight":1.0, "countIfUndefined":"false", "params": {}}], "threshold":1.0, "aggregation": "MIN", "positive":"layer3", "negative":"NO_MATCH", "undefined":"layer3", "ignoreUndefined": "false"},
|
|
"layer3": {"fields": [{"field":"title", "comparator":"JaroWinkler", "weight":1.0, "countIfUndefined":"false", "params":{}}], "threshold": 0.99, "aggregation": "MAX", "positive":"MATCH", "negative":"NO_MATCH", "undefined":"NO_MATCH", "ignoreUndefined": "false"}
|
|
},
|
|
"model" : [
|
|
{ "name" : "pid", "type" : "String", "path" : "pid[qualifier#classid = {doi}]/value", "overrideMatch" : "true" },
|
|
{ "name" : "title", "type" : "String", "path" : "result/metadata/title[qualifier#classid = {main title}]/value" },
|
|
{ "name" : "dateofacceptance", "type" : "String", "path" : "result/metadata/dateofacceptance/value" } ,
|
|
{ "name" : "authors", "type" : "List", "path" : "result/author/metadata/fullname/value" }
|
|
],
|
|
"blacklists" : {
|
|
"title" : [
|
|
"^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
|
|
"^(Kiri Karl Morgensternile).*$",
|
|
"^(\\[Eksliibris Aleksandr).*\\]$",
|
|
"^(\\[Eksliibris Aleksandr).*$",
|
|
"^(Eksliibris Aleksandr).*$",
|
|
"^(Kiri A\\. de Vignolles).*$",
|
|
"^(2 kirja Karl Morgensternile).*$",
|
|
"^(Pirita kloostri idaosa arheoloogilised).*$",
|
|
"^(Kiri tundmatule).*$",
|
|
"^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
|
|
"^(Eksliibris Nikolai Birukovile).*$",
|
|
"^(Eksliibris Nikolai Issakovile).*$",
|
|
"^(WHP Cruise Summary Information of section).*$",
|
|
"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
|
|
"^(Measurement of the spin\\-dependent structure function).*"
|
|
] },
|
|
"synonyms": {}
|
|
}
|
|
|
|
}
|