Project

General

Profile

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", "resultProject_outcome_isProducedBy", "resultResult_publicationDataset_isRelatedTo", "resultResult_similarity_isAmongTopNSimilarDocuments", "resultResult_similarity_hasAmongTopNSimilarDocuments" ],
11
    "includeChildren" : "true",
12
    "maxChildren" : "40"
13
  },
14
  "pace" : {
15
    "clustering" : [
16
      { "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
17
      { "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } },
18
      { "name" : "lowercase", "fields" : [ "doi" ], "params" : { } }
19
    ],
20
    "strictConditions" : [
21
      { "name" : "pidMatch", "fields" : [ "pid" ] }
22
    ],
23
    "conditions" : [
24
      { "name" : "titleVersionMatch", "fields" : [ "title" ] },
25
      { "name" : "sizeMatch", "fields" : [ "authors" ] }
26
    ],
27
    "model" : [
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" },
30
      { "name" : "title", "algo" : "LevensteinTitle", "type" : "String", "weight" : "1.0", "ignoreMissing" : "false", "path" : "result/metadata/title[qualifier#classid = {main title}]/value" },
31
      { "name" : "authors", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/metadata/author/fullname" }
32
    ],
33
    "blacklists" : {
34
      "title" : [
35
        "^Inside Front Cover$",
36
        "(?i)^Poster presentations$",
37
        "^THE ASSOCIATION AND THE GENERAL MEDICAL COUNCIL$",
38
        "^Problems with perinatal pathology\.?$",
39
        "(?i)^Cases? of Puerperal Convulsions$",
40
        "(?i)^Operative Gyna?ecology$",
41
        "(?i)^Mind the gap\!?\:?$",
42
        "^Chronic fatigue syndrome\.?$",
43
        "^Cartas? ao editor Letters? to the Editor$",
44
        "^Note from the Editor$",
45
        "^Anesthesia Abstract$",
46

    
47
        "^Annual report$",
48
        "(?i)^“?THE RADICAL PREVENTION OF VENEREAL DISEASE\.?”?$",
49
        "(?i)^Graph and Table of Infectious Diseases?$",
50
        "^Presentation$",
51
        "(?i)^Reviews and Information on Publications$",
52
        "(?i)^PUBLIC HEALTH SERVICES?$",
53
        "(?i)^COMBINED TEXT-?BOOK OF OBSTETRICS AND GYN(Æ|ae)COLOGY$",
54
        "(?i)^Adrese autora$",
55
        "(?i)^Systematic Part .*\. Catalogus Fossilium Austriae, Band 2: Echinoidea neogenica$",
56
        "(?i)^Acknowledgement to Referees$",
57
        "(?i)^Behçet's disease\.?$",
58
        "(?i)^Isolation and identification of restriction endonuclease.*$",
59
        "(?i)^CEREBROVASCULAR DISEASES?.?$",
60
        "(?i)^Screening for abdominal aortic aneurysms?\.?$",
61
        "^Event management$",
62
        "(?i)^Breakfast and Crohn's disease.*\.?$",
63
        "^Cálculo de concentraciones en disoluciones acuosas. Ejercicio interactivo\..*\.$",
64
        "(?i)^Genetic and functional analyses of SHANK2 mutations suggest a multiple hit model of Autism spectrum disorders?\.?$",
65
        "^Gushi hakubutsugaku$",
66

    
67
        "^Starobosanski nadpisi u Bosni i Hercegovini \(.*\)$",
68
        "^Intestinal spirocha?etosis$",
69
        "^Treatment of Rodent Ulcer$",
70
        "(?i)^\W*Cloud Computing\W*$",
71
        "^Compendio mathematico : en que se contienen todas las materias mas principales de las Ciencias que tratan de la cantidad$",
72
        "^Free Communications, Poster Presentations: Session [A-F]$",
73

    
74
        "^“The Historical Aspects? of Quackery\.?”$",
75
        "^A designated centre for people with disabilities operated by St John of God Community Services (Limited|Ltd), Louth$",
76
        "^P(er|re)-Mile Premiums for Auto Insurance\\.?$",
77
        "(?i)^Case Report$",
78
        "^Boletín Informativo$",
79
        "(?i)^Glioblastoma Multiforme$",
80
        "(?i)^Nuevos táxones animales descritos en la península Ibérica y Macaronesia desde 1994 \\(.*\\)$",
81
        "^Zaměstnanecké výhody$",
82
        "(?i)^The Economics of Terrorism and Counter-Terrorism: A Survey \\(Part .*\\)$",
83
        "(?i)^Carotid body tumours?\\.?$",
84
        "(?i)^\\[Españoles en Francia : La condición Emigrante.*\\]$",
85
        "^Avant-propos$",
86
        "(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Head(s)? and Capital(s)?$",
87
        "(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Bases?$",
88
        "(?i)^PUBLIC HEALTH VERSUS THE STATE$",
89
        "^Viñetas de Cortázar$",
90
        "(?i)^Search for heavy neutrinos and W(\\[|_|\\(|_\\{|-)?R(\\]|\\)|\\})? bosons with right-handed couplings in a left-right symmetric model in pp collisions at.*TeV(\.)?$",
91
        "(?i)^Measurement of the pseudorapidity and centrality dependence of the transverse energy density in Pb(-?)Pb collisions at.*tev(\.?)$",
92
        "(?i)^Search for resonances decaying into top-quark pairs using fully hadronic decays in pp collisions with ATLAS at.*TeV$",
93
        "(?i)^Search for neutral minimal supersymmetric standard model Higgs bosons decaying to tau pairs in pp collisions at.*tev$",
94

    
95
        "(?i)^Relatório de Estágio (de|em) Angiologia e Cirurgia Vascular$",
96
        "^Aus der AGMB$",
97

    
98
        "^Znanstveno-stručni prilozi$",
99
        "(?i)^Zhodnocení finanční situace podniku a návrhy na zlepšení$",
100
        "(?i)^Evaluation of the Financial Situation in the Firm and Proposals to its Improvement$",
101
        "(?i)^Hodnocení finanční situace podniku a návrhy na její zlepšení$",
102
        "^Finanční analýza podniku$",
103
        "^Financial analysis( of business)?$",
104
        "(?i)^Textbook of Gyn(a)?(Æ)?(e)?cology$",
105
        "^Jikken nihon shūshinsho$",
106
        "(?i)^CORONER('|s)(s|') INQUESTS$",
107
        "(?i)^(Μελέτη παραγόντων )?risk management( για ανάπτυξη και εφαρμογή ενός πληροφοριακού συστήματος| και ανάπτυξη συστήματος)?$",
108
        "(?i)^Consultants' contract(s)?$",
109
        "(?i)^Upute autorima$",
110
        "(?i)^Bijdrage tot de Kennis van den Godsdienst der Dajaks van Lan(d|f)ak en Tajan$",
111
        "^Joshi shin kokubun$",
112
        "^Kōtō shōgaku dokuhon nōson'yō$",
113
        "^Jinjō shōgaku shōka$",
114
        "^Shōgaku shūjichō$",
115
        "^Nihon joshi dokuhon$",
116
        "^Joshi shin dokuhon$",
117
        "^Chūtō kanbun dokuhon$",
118
        "^Wabun dokuhon$",
119
        "(?i)^(Analysis of economy selected village or town|Rozbor hospodaření vybrané obce či města)$",
120
        "(?i)^cardiac rehabilitation$",
121
        "(?i)^Analytical summary$",
122
        "^Thesaurus resolutionum Sacrae Congregationis Concilii$",
123
        "(?i)^Sumario analítico(\\s{1})?(Analitic summary)?$",
124
        "^Prikazi i osvrti$",
125
        "^Rodinný dům s provozovnou$",
126
        "^Family house with an establishment$",
127
        "^Shinsei chūtō shin kokugun$",
128
        "^Pulmonary alveolar proteinosis(\\.?)$",
129
        "^Shinshū kanbun$",
130
        "^Viñeta(s?) de Rodríguez$",
131
        "(?i)^RUBRIKA UREDNIKA$",
132
        "^A Matching Model of the Academic Publication Market$",
133
        "^Yōgaku kōyō$",
134

    
135
        "^Internetový marketing$",
136
        "^Internet marketing$",
137
        "^Chūtō kokugo dokuhon$",
138
        "^Kokugo dokuhon$",
139
        "^Antibiotic Cover for Dental Extraction(s?)$",
140
        "^Strategie podniku$",
141
        "^Strategy of an Enterprise$",
142
        "(?i)^respiratory disease(s?)(\.?)$",
143
        "^Award(s?) for Gallantry in Civil Defence$",
144
        "^Podniková kultura$",
145
        "^Corporate Culture$",
146
        "^Severe hyponatraemia in hospital inpatient(s?)(\.?)$",
147
        "^Pracovní motivace$",
148
        "^Work Motivation$",
149
        "^Kaitei kōtō jogaku dokuhon$",
150
        "^Konsolidovaná účetní závěrka$",
151
        "^Consolidated Financial Statements$",
152
        "(?i)^intracranial tumour(s?)$",
153
        "^Climate Change Mitigation Options and Directed Technical Change: A Decentralized Equilibrium Analysis$",
154
        "^\\[CERVECERIAS MAHOU(\\.|\\:) INTERIOR\\] \\[Material gráfico\\]$",
155
        "^Housing Market Dynamics(\\:|\\.) On the Contribution of Income Shocks and Credit Constraint(s?)$",
156
        "^\\[Funciones auxiliares de la música en Radio París,.*\\]$",
157
        "^Úroveň motivačního procesu jako způsobu vedení lidí$",
158
        "^The level of motivation process as a leadership$",
159
        "^Pay-beds in N(\\.?)H(\\.?)S(\\.?) Hospitals$",
160
        "(?i)^news and events$",
161
        "(?i)^NOVOSTI I DOGAĐAJI$",
162
        "^Sansū no gakushū$",
163
        "^Posouzení informačního systému firmy a návrh změn$",
164
        "^Information System Assessment and Proposal for ICT Modification$",
165
        "^Stresové zatížení pracovníků ve vybrané profesi$",
166
        "^Stress load in a specific job$",
167

    
168
        "^Sunday: Poster Sessions, Pt.*$",
169
        "^Monday: Poster Sessions, Pt.*$",
170
        "^Wednesday: Poster Sessions, Pt.*",
171
        "^Tuesday: Poster Sessions, Pt.*$",
172

    
173
        "^Analýza reklamy$",
174
        "^Analysis of advertising$",
175

    
176
        "^Shōgaku shūshinsho$",
177
        "^Shōgaku sansū$",
178
        "^Shintei joshi kokubun$",
179
        "^Taishō joshi kokubun dokuhon$",
180
        "^Joshi kokubun$",
181

    
182
        "^Účetní uzávěrka a účetní závěrka v ČR$",
183
        "(?i)^The \"?Causes\"? of Cancer$",
184
        "^Normas para la publicación de artículos$",
185
        "^Editor('|s)(s|') [Rr]eply$",
186
        "^Editor(’|s)(s|’) letter$",
187
        "^Redaktoriaus žodis$",
188
        "^DISCUSSION ON THE PRECEDING PAPER$",
189
        "^Kōtō shōgaku shūshinsho jidōyō$",
190
        "^Shōgaku nihon rekishi$",
191
        "^(Theory of the flow of action currents in isolated myelinated nerve fibers).*$",
192
        "^Préface$",
193
        "^Occupational [Hh]ealth [Ss]ervices.$",
194
        "^In Memoriam Professor Toshiyuki TAKESHIMA$",
195
        "^Účetní závěrka ve vybraném podniku.*$",
196
        "^Financial statements in selected company$",
197
        "^Abdominal [Aa]ortic [Aa]neurysms.*$",
198
        "^Pseudomyxoma peritonei$",
199
        "^Kazalo autora$",
200

    
201
        "(?i)^uvodna riječ$",
202
        "^Motivace jako způsob vedení lidí$",
203
        "^Motivation as a leadership$",
204
        "^Polyfunkční dům$",
205
        "^Multi\\-funkcional building$",
206
        "^Podnikatelský plán$",
207
        "(?i)^Podnikatelský záměr$",
208
        "(?i)^Business Plan$",
209
        "^Oceňování nemovitostí$",
210
        "^Marketingová komunikace$",
211
        "^Marketing communication$",
212
        "^Sumario Analítico$",
213
        "^Riječ uredništva$",
214
        "^Savjetovanja i priredbe$",
215
        "^Índice$",
216
        "^(Starobosanski nadpisi).*$",
217
        "^Vzdělávání pracovníků v organizaci$",
218
        "^Staff training in organization$",
219
        "^(Life Histories of North American Geometridae).*$",
220
        "^Strategická analýza podniku$",
221
        "^Strategic Analysis of an Enterprise$",
222
        "^Sadržaj$",
223
        "^Upute suradnicima$",
224
        "^Rodinný dům$",
225
        "(?i)^Fami(l)?ly house$",
226
        "^Upute autorima$",
227
        "^Strategic Analysis$",
228
        "^Finanční analýza vybraného podniku$",
229
        "^Finanční analýza$",
230
        "^Riječ urednika$",
231
        "(?i)^Content(s?)$",
232
        "(?i)^Inhalt$",
233
        "^Jinjō shōgaku shūshinsho jidōyō$",
234
        "(?i)^Index$",
235
        "^Chūgaku kokubun kyōkasho$",
236
        "^Retrato de una mujer$",
237
        "^Retrato de un hombre$",
238
        "^Kōtō shōgaku dokuhon$",
239
        "^Shotōka kokugo$",
240
        "^Shōgaku dokuhon$",
241
        "^Jinjō shōgaku kokugo dokuhon$",
242
        "^Shinsei kokugo dokuhon$",
243
        "^Teikoku dokuhon$",
244
        "^Instructions to Authors$",
245
        "^KİTAP TAHLİLİ$",
246
        "^PRZEGLĄD PIŚMIENNICTWA$",
247
        "(?i)^Presentación$",
248
        "^İçindekiler$",
249
        "(?i)^Tabl?e of contents$",
250
        "^(CODICE DEL BEATO DE LOS REYES FERNANDO I Y SANCHA).*$",
251
        "^(\\[MADRID\\. BIBL\\. NAC\\. N.*KING FERDINAND I.*FROM SAN ISIDORO DE LEON\\. FACUNDUS SCRIPSIT DATED.*\\]).*",
252
        "^Editorial( Board)?$",
253
        "(?i)^Editorial \\(English\\)$",
254
        "^Editörden$",
255
        "^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
256
        "^(Kiri Karl Morgensternile).*$",
257
        "^(\\[Eksliibris Aleksandr).*\\]$",
258
        "^(\\[Eksliibris Aleksandr).*$",
259
        "^(Eksliibris Aleksandr).*$",
260
        "^(Kiri A\\. de Vignolles).*$",
261
        "^(2 kirja Karl Morgensternile).*$",
262
        "^(Pirita kloostri idaosa arheoloogilised).*$",
263
        "^(Kiri tundmatule).*$",
264
        "^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
265
        "^(Eksliibris Nikolai Birukovile).*$",
266
        "^(Eksliibris Nikolai Issakovile).*$",
267
        "^(WHP Cruise Summary Information of section).*$",
268
        "^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
269
        "^(Measurement of the spin\\-dependent structure function).*",
270
        "(?i)^.*authors['’′]? reply\.?$",
271
        "(?i)^.*authors['’′]? response\.?$"
272
      ]
273
    }
274
  }
275
}
(8-8/9)