Revision 50230
Added by Claudio Atzori almost 6 years ago
| result.prod.pace.conf | ||
|---|---|---|
| 1 |
{
|
|
| 2 |
"wf" : {
|
|
| 3 |
"threshold" : "0.99", |
|
| 4 |
"dedupRun" : "001", |
|
| 5 |
"entityType" : "result", |
|
| 6 |
"orderField" : "title", |
|
| 7 |
"queueMaxSize" : "2000", |
|
| 8 |
"groupMaxSize" : "10", |
|
| 9 |
"slidingWindowSize" : "200", |
|
| 10 |
"rootBuilder" : [ "result", "resultProject_outcome_isProducedBy", "resultResult_publicationDataset_isRelatedTo", "resultResult_similarity_isAmongTopNSimilarDocuments", "resultResult_similarity_hasAmongTopNSimilarDocuments" ], |
|
| 11 |
"includeChildren" : "true" |
|
| 12 |
}, |
|
| 13 |
"pace" : {
|
|
| 1 |
{
|
|
| 2 |
"wf" : {
|
|
| 3 |
"threshold" : "0.99", |
|
| 4 |
"dedupRun" : "001", |
|
| 5 |
"entityType" : "result", |
|
| 6 |
"orderField" : "title", |
|
| 7 |
"queueMaxSize" : "4000", |
|
| 8 |
"groupMaxSize" : "40", |
|
| 9 |
"slidingWindowSize" : "200", |
|
| 10 |
"rootBuilder" : [ "result", "personResult_authorship_hasAuthor", "resultProject_outcome_isProducedBy", "resultResult_publicationDataset_isRelatedTo", "resultResult_similarity_isAmongTopNSimilarDocuments", "resultResult_similarity_hasAmongTopNSimilarDocuments" ], |
|
| 11 |
"includeChildren" : "true", |
|
| 12 |
"maxChildren" : "40" |
|
| 13 |
}, |
|
| 14 |
"pace" : {
|
|
| 14 | 15 |
"clustering" : [ |
| 15 | 16 |
{ "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
|
| 16 |
{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } }
|
|
| 17 |
{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } },
|
|
| 18 |
{ "name" : "lowercase", "fields" : [ "doi" ], "params" : { } }
|
|
| 17 | 19 |
], |
| 18 | 20 |
"strictConditions" : [ |
| 19 |
{ "name" : "exactMatch", "fields" : [ "pid" ] }
|
|
| 20 |
],
|
|
| 21 |
"conditions" : [
|
|
| 22 |
{ "name" : "titleVersionMatch", "fields" : [ "title" ] },
|
|
| 23 |
{ "name" : "sizeMatch", "fields" : [ "authors" ] }
|
|
| 24 |
],
|
|
| 21 |
{ "name" : "pidMatch", "fields" : [ "pid" ] }
|
|
| 22 |
], |
|
| 23 |
"conditions" : [
|
|
| 24 |
{ "name" : "titleVersionMatch", "fields" : [ "title" ] },
|
|
| 25 |
{ "name" : "sizeMatch", "fields" : [ "authors" ] }
|
|
| 26 |
],
|
|
| 25 | 27 |
"model" : [ |
| 26 |
{ "name" : "pid", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid[qualifier#classid = {doi}]/value", "overrideMatch" : "true" },
|
|
| 28 |
{ "name" : "doi", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid[qualifier#classid = {doi}]/value" },
|
|
| 29 |
{ "name" : "pid", "algo" : "Null", "type" : "JSON", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid", "overrideMatch" : "true" },
|
|
| 27 | 30 |
{ "name" : "title", "algo" : "LevensteinTitle", "type" : "String", "weight" : "1.0", "ignoreMissing" : "false", "path" : "result/metadata/title[qualifier#classid = {main title}]/value" },
|
| 28 |
{ "name" : "authors", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/metadata/author/fullname" }
|
|
| 31 |
{ "name" : "authors", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/author/metadata/fullname/value" }
|
|
| 29 | 32 |
], |
| 30 |
"blacklists" : {
|
|
| 33 |
"blacklists" : {
|
|
| 31 | 34 |
"title" : [ |
| 32 | 35 |
"^THE ASSOCIATION AND THE GENERAL MEDICAL COUNCIL$", |
| 33 | 36 |
"^Problems with perinatal pathology\.?$", |
| ... | ... | |
| 38 | 41 |
"^Cartas? ao editor Letters? to the Editor$", |
| 39 | 42 |
"^Note from the Editor$", |
| 40 | 43 |
"^Anesthesia Abstract$", |
| 41 |
|
|
| 44 |
|
|
| 42 | 45 |
"^Annual report$", |
| 43 | 46 |
"(?i)^“?THE RADICAL PREVENTION OF VENEREAL DISEASE\.?”?$", |
| 44 | 47 |
"(?i)^Graph and Table of Infectious Diseases?$", |
| ... | ... | |
| 58 | 61 |
"^Cálculo de concentraciones en disoluciones acuosas. Ejercicio interactivo\..*\.$", |
| 59 | 62 |
"(?i)^Genetic and functional analyses of SHANK2 mutations suggest a multiple hit model of Autism spectrum disorders?\.?$", |
| 60 | 63 |
"^Gushi hakubutsugaku$", |
| 61 |
|
|
| 62 |
"^Starobosanski nadpisi u Bosni i Hercegovini \(.*\)$",
|
|
| 64 |
|
|
| 65 |
"^Starobosanski nadpisi u Bosni i Hercegovini \(.*\)$", |
|
| 63 | 66 |
"^Intestinal spirocha?etosis$", |
| 64 | 67 |
"^Treatment of Rodent Ulcer$", |
| 65 | 68 |
"(?i)^\W*Cloud Computing\W*$", |
| 66 |
"^Compendio mathematico : en que se contienen todas las materias mas principales de las Ciencias que tratan de la cantidad$",
|
|
| 69 |
"^Compendio mathematico : en que se contienen todas las materias mas principales de las Ciencias que tratan de la cantidad$", |
|
| 67 | 70 |
"^Free Communications, Poster Presentations: Session [A-F]$", |
| 68 |
|
|
| 71 |
|
|
| 69 | 72 |
"^“The Historical Aspects? of Quackery\.?”$", |
| 70 | 73 |
"^A designated centre for people with disabilities operated by St John of God Community Services (Limited|Ltd), Louth$", |
| 71 | 74 |
"^P(er|re)-Mile Premiums for Auto Insurance\\.?$", |
| 72 |
"(?i)^Case Report$",
|
|
| 75 |
"(?i)^Case Report$", |
|
| 73 | 76 |
"^Boletín Informativo$", |
| 74 | 77 |
"(?i)^Glioblastoma Multiforme$", |
| 75 | 78 |
"(?i)^Nuevos táxones animales descritos en la península Ibérica y Macaronesia desde 1994 \\(.*\\)$", |
| 76 | 79 |
"^Zaměstnanecké výhody$", |
| 77 | 80 |
"(?i)^The Economics of Terrorism and Counter-Terrorism: A Survey \\(Part .*\\)$", |
| 78 |
"(?i)^Carotid body tumours?\\.?$",
|
|
| 81 |
"(?i)^Carotid body tumours?\\.?$", |
|
| 79 | 82 |
"(?i)^\\[Españoles en Francia : La condición Emigrante.*\\]$", |
| 80 | 83 |
"^Avant-propos$", |
| 81 | 84 |
"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Head(s)? and Capital(s)?$", |
| 82 | 85 |
"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Bases?$", |
| 83 |
"(?i)^PUBLIC HEALTH VERSUS THE STATE$",
|
|
| 86 |
"(?i)^PUBLIC HEALTH VERSUS THE STATE$", |
|
| 84 | 87 |
"^Viñetas de Cortázar$", |
| 85 | 88 |
"(?i)^Search for heavy neutrinos and W(\\[|_|\\(|_\\{|-)?R(\\]|\\)|\\})? bosons with right-handed couplings in a left-right symmetric model in pp collisions at.*TeV(\.)?$",
|
| 86 |
"(?i)^Measurement of the pseudorapidity and centrality dependence of the transverse energy density in Pb(-?)Pb collisions at.*tev(\.?)$",
|
|
| 89 |
"(?i)^Measurement of the pseudorapidity and centrality dependence of the transverse energy density in Pb(-?)Pb collisions at.*tev(\.?)$", |
|
| 87 | 90 |
"(?i)^Search for resonances decaying into top-quark pairs using fully hadronic decays in pp collisions with ATLAS at.*TeV$", |
| 88 | 91 |
"(?i)^Search for neutral minimal supersymmetric standard model Higgs bosons decaying to tau pairs in pp collisions at.*tev$", |
| 89 |
|
|
| 92 |
|
|
| 90 | 93 |
"(?i)^Relatório de Estágio (de|em) Angiologia e Cirurgia Vascular$", |
| 91 |
"^Aus der AGMB$",
|
|
| 92 |
|
|
| 94 |
"^Aus der AGMB$", |
|
| 95 |
|
|
| 93 | 96 |
"^Znanstveno-stručni prilozi$", |
| 94 |
"^Zhodnocení finanční situace podniku a návrhy na zlepšení$", |
|
| 95 |
"^Evaluation of the Financial Situation in the Firm and Proposals to its Improvement$", |
|
| 97 |
"(?i)^Zhodnocení finanční situace podniku a návrhy na zlepšení$", |
|
| 98 |
"(?i)^Evaluation of the Financial Situation in the Firm and Proposals to its Improvement$", |
|
| 99 |
"(?i)^Hodnocení finanční situace podniku a návrhy na její zlepšení$", |
|
| 96 | 100 |
"^Finanční analýza podniku$", |
| 97 | 101 |
"^Financial analysis( of business)?$", |
| 98 | 102 |
"(?i)^Textbook of Gyn(a)?(Æ)?(e)?cology$", |
| 99 | 103 |
"^Jikken nihon shūshinsho$", |
| 100 | 104 |
"(?i)^CORONER('|s)(s|') INQUESTS$",
|
| 101 |
"(?i)^(Μελέτη παραγόντων )?risk management( για ανάπτυξη και εφαρμογή ενός πληροφοριακού συστήματος| και ανάπτυξη συστήματος)?$",
|
|
| 105 |
"(?i)^(Μελέτη παραγόντων )?risk management( για ανάπτυξη και εφαρμογή ενός πληροφοριακού συστήματος| και ανάπτυξη συστήματος)?$", |
|
| 102 | 106 |
"(?i)^Consultants' contract(s)?$", |
| 103 | 107 |
"(?i)^Upute autorima$", |
| 104 | 108 |
"(?i)^Bijdrage tot de Kennis van den Godsdienst der Dajaks van Lan(d|f)ak en Tajan$", |
| ... | ... | |
| 114 | 118 |
"(?i)^cardiac rehabilitation$", |
| 115 | 119 |
"(?i)^Analytical summary$", |
| 116 | 120 |
"^Thesaurus resolutionum Sacrae Congregationis Concilii$", |
| 117 |
"(?i)^Sumario analítico(\\s{1})?(Analitic summary)?$",
|
|
| 121 |
"(?i)^Sumario analítico(\\s{1})?(Analitic summary)?$",
|
|
| 118 | 122 |
"^Prikazi i osvrti$", |
| 119 | 123 |
"^Rodinný dům s provozovnou$", |
| 120 | 124 |
"^Family house with an establishment$", |
| ... | ... | |
| 125 | 129 |
"(?i)^RUBRIKA UREDNIKA$", |
| 126 | 130 |
"^A Matching Model of the Academic Publication Market$", |
| 127 | 131 |
"^Yōgaku kōyō$", |
| 128 |
|
|
| 132 |
|
|
| 129 | 133 |
"^Internetový marketing$", |
| 130 | 134 |
"^Internet marketing$", |
| 131 | 135 |
"^Chūtō kokugo dokuhon$", |
| 132 | 136 |
"^Kokugo dokuhon$", |
| 133 | 137 |
"^Antibiotic Cover for Dental Extraction(s?)$", |
| 134 |
"^Strategie podniku$",
|
|
| 138 |
"^Strategie podniku$", |
|
| 135 | 139 |
"^Strategy of an Enterprise$", |
| 136 | 140 |
"(?i)^respiratory disease(s?)(\.?)$", |
| 137 | 141 |
"^Award(s?) for Gallantry in Civil Defence$", |
| ... | ... | |
| 158 | 162 |
"^Information System Assessment and Proposal for ICT Modification$", |
| 159 | 163 |
"^Stresové zatížení pracovníků ve vybrané profesi$", |
| 160 | 164 |
"^Stress load in a specific job$", |
| 161 |
|
|
| 165 |
|
|
| 162 | 166 |
"^Sunday: Poster Sessions, Pt.*$", |
| 163 | 167 |
"^Monday: Poster Sessions, Pt.*$", |
| 164 | 168 |
"^Wednesday: Poster Sessions, Pt.*", |
| 165 | 169 |
"^Tuesday: Poster Sessions, Pt.*$", |
| 166 |
|
|
| 170 |
|
|
| 167 | 171 |
"^Analýza reklamy$", |
| 168 | 172 |
"^Analysis of advertising$", |
| 169 |
|
|
| 173 |
|
|
| 170 | 174 |
"^Shōgaku shūshinsho$", |
| 171 | 175 |
"^Shōgaku sansū$", |
| 172 | 176 |
"^Shintei joshi kokubun$", |
| 173 | 177 |
"^Taishō joshi kokubun dokuhon$", |
| 174 |
"^Joshi kokubun$",
|
|
| 175 |
|
|
| 178 |
"^Joshi kokubun$", |
|
| 179 |
|
|
| 176 | 180 |
"^Účetní uzávěrka a účetní závěrka v ČR$", |
| 177 | 181 |
"(?i)^The \"?Causes\"? of Cancer$", |
| 178 | 182 |
"^Normas para la publicación de artículos$", |
| 179 | 183 |
"^Editor('|s)(s|') [Rr]eply$",
|
| 180 | 184 |
"^Editor(’|s)(s|’) letter$", |
| 181 |
"^Redaktoriaus žodis$",
|
|
| 185 |
"^Redaktoriaus žodis$", |
|
| 182 | 186 |
"^DISCUSSION ON THE PRECEDING PAPER$", |
| 183 | 187 |
"^Kōtō shōgaku shūshinsho jidōyō$", |
| 184 | 188 |
"^Shōgaku nihon rekishi$", |
| ... | ... | |
| 190 | 194 |
"^Financial statements in selected company$", |
| 191 | 195 |
"^Abdominal [Aa]ortic [Aa]neurysms.*$", |
| 192 | 196 |
"^Pseudomyxoma peritonei$", |
| 193 |
"^Kazalo autora$",
|
|
| 194 |
|
|
| 197 |
"^Kazalo autora$", |
|
| 198 |
|
|
| 195 | 199 |
"(?i)^uvodna riječ$", |
| 196 | 200 |
"^Motivace jako způsob vedení lidí$", |
| 197 | 201 |
"^Motivation as a leadership$", |
| 198 | 202 |
"^Polyfunkční dům$", |
| 199 | 203 |
"^Multi\\-funkcional building$", |
| 200 | 204 |
"^Podnikatelský plán$", |
| 201 |
"^Business Plan$", |
|
| 205 |
"(?i)^Podnikatelský záměr$", |
|
| 206 |
"(?i)^Business Plan$", |
|
| 202 | 207 |
"^Oceňování nemovitostí$", |
| 203 | 208 |
"^Marketingová komunikace$", |
| 204 | 209 |
"^Marketing communication$", |
| ... | ... | |
| 244 | 249 |
"^(\\[MADRID\\. BIBL\\. NAC\\. N.*KING FERDINAND I.*FROM SAN ISIDORO DE LEON\\. FACUNDUS SCRIPSIT DATED.*\\]).*", |
| 245 | 250 |
"^Editorial( Board)?$", |
| 246 | 251 |
"(?i)^Editorial \\(English\\)$", |
| 247 |
"^Editörden$",
|
|
| 252 |
"^Editörden$", |
|
| 248 | 253 |
"^(Corpus Oral Dialectal \\(COD\\)\\.).*$", |
| 249 | 254 |
"^(Kiri Karl Morgensternile).*$", |
| 250 | 255 |
"^(\\[Eksliibris Aleksandr).*\\]$", |
| ... | ... | |
| 259 | 264 |
"^(Eksliibris Nikolai Issakovile).*$", |
| 260 | 265 |
"^(WHP Cruise Summary Information of section).*$", |
| 261 | 266 |
"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$", |
| 262 |
"^(Measurement of the spin\\-dependent structure function).*" |
|
| 267 |
"^(Measurement of the spin\\-dependent structure function).*", |
|
| 268 |
"(?i)^.*authors['’′]? reply\.?$", |
|
| 269 |
"(?i)^.*authors['’′]? response\.?$" |
|
| 263 | 270 |
] |
| 264 |
}
|
|
| 271 |
} |
|
| 265 | 272 |
} |
| 266 | 273 |
} |
Also available in: Unified diff
aligned with the configuration on production