Project

General

Profile

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" : {		
14
		"clustering" : [
15
			{ "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
16
			{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } } 
17
		],
18
		"strictConditions" : [
19
			{ "name" : "exactMatch", "fields" : [ "pid" ] }
20
		], 
21
  		"conditions" : [ 
22
  			{ "name" : "titleVersionMatch", "fields" : [ "title" ] },
23
  			{ "name" : "sizeMatch", "fields" : [ "authors" ] }
24
  		],		
25
		"model" : [
26
			{ "name" : "pid", "algo" : "Null", "type" : "String", "weight" : "0.0", "ignoreMissing" : "true", "path" : "pid[qualifier#classid = {doi}]/value", "overrideMatch" : "true" }, 	
27
			{ "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" }
29
		],
30
		"blacklists" : { 
31
			"title" : [
32
				"^THE ASSOCIATION AND THE GENERAL MEDICAL COUNCIL$",
33
				"^Problems with perinatal pathology\.?$",
34
				"(?i)^Cases? of Puerperal Convulsions$",
35
				"(?i)^Operative Gyna?ecology$",
36
				"(?i)^Mind the gap\!?\:?$",
37
				"^Chronic fatigue syndrome\.?$",
38
				"^Cartas? ao editor Letters? to the Editor$",
39
				"^Note from the Editor$",
40
				"^Anesthesia Abstract$",
41
				
42
				"^Annual report$",
43
				"(?i)^“?THE RADICAL PREVENTION OF VENEREAL DISEASE\.?”?$",
44
				"(?i)^Graph and Table of Infectious Diseases?$",
45
				"^Presentation$",
46
				"(?i)^Reviews and Information on Publications$",
47
				"(?i)^PUBLIC HEALTH SERVICES?$",
48
				"(?i)^COMBINED TEXT-?BOOK OF OBSTETRICS AND GYN(Æ|ae)COLOGY$",
49
				"(?i)^Adrese autora$",
50
				"(?i)^Systematic Part .*\. Catalogus Fossilium Austriae, Band 2: Echinoidea neogenica$",
51
				"(?i)^Acknowledgement to Referees$",
52
				"(?i)^Behçet's disease\.?$",
53
				"(?i)^Isolation and identification of restriction endonuclease.*$",
54
				"(?i)^CEREBROVASCULAR DISEASES?.?$",
55
				"(?i)^Screening for abdominal aortic aneurysms?\.?$",
56
				"^Event management$",
57
				"(?i)^Breakfast and Crohn's disease.*\.?$",
58
				"^Cálculo de concentraciones en disoluciones acuosas. Ejercicio interactivo\..*\.$",
59
				"(?i)^Genetic and functional analyses of SHANK2 mutations suggest a multiple hit model of Autism spectrum disorders?\.?$",
60
				"^Gushi hakubutsugaku$",
61
			
62
				"^Starobosanski nadpisi u Bosni i Hercegovini \(.*\)$",							
63
				"^Intestinal spirocha?etosis$",
64
				"^Treatment of Rodent Ulcer$",
65
				"(?i)^\W*Cloud Computing\W*$",
66
				"^Compendio mathematico : en que se contienen todas las materias mas principales de las Ciencias que tratan de la cantidad$",				
67
				"^Free Communications, Poster Presentations: Session [A-F]$",
68
				
69
				"^“The Historical Aspects? of Quackery\.?”$",
70
				"^A designated centre for people with disabilities operated by St John of God Community Services (Limited|Ltd), Louth$",
71
				"^P(er|re)-Mile Premiums for Auto Insurance\\.?$",
72
				"(?i)^Case Report$",							
73
				"^Boletín Informativo$",
74
				"(?i)^Glioblastoma Multiforme$",
75
				"(?i)^Nuevos táxones animales descritos en la península Ibérica y Macaronesia desde 1994 \\(.*\\)$",
76
				"^Zaměstnanecké výhody$",
77
				"(?i)^The Economics of Terrorism and Counter-Terrorism: A Survey \\(Part .*\\)$",
78
				"(?i)^Carotid body tumours?\\.?$", 
79
				"(?i)^\\[Españoles en Francia : La condición Emigrante.*\\]$",
80
				"^Avant-propos$",
81
				"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Head(s)? and Capital(s)?$",
82
				"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Bases?$",
83
				"(?i)^PUBLIC HEALTH VERSUS THE STATE$",							
84
				"^Viñetas de Cortázar$",
85
				"(?i)^Search for heavy neutrinos and W(\\[|_|\\(|_\\{|-)?R(\\]|\\)|\\})? bosons with right-handed couplings in a left-right symmetric model in pp collisions at.*TeV(\.)?$",
86
				"(?i)^Measurement of the pseudorapidity and centrality dependence of the transverse energy density in Pb(-?)Pb collisions at.*tev(\.?)$",				
87
				"(?i)^Search for resonances decaying into top-quark pairs using fully hadronic decays in pp collisions with ATLAS at.*TeV$",
88
				"(?i)^Search for neutral minimal supersymmetric standard model Higgs bosons decaying to tau pairs in pp collisions at.*tev$",
89
			
90
				"(?i)^Relatório de Estágio (de|em) Angiologia e Cirurgia Vascular$",
91
				"^Aus der AGMB$",				
92
			
93
				"^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$",
96
				"^Finanční analýza podniku$",
97
				"^Financial analysis( of business)?$",
98
				"(?i)^Textbook of Gyn(a)?(Æ)?(e)?cology$",
99
				"^Jikken nihon shūshinsho$",
100
				"(?i)^CORONER('|s)(s|') INQUESTS$",
101
				"(?i)^(Μελέτη παραγόντων )?risk management( για ανάπτυξη και εφαρμογή ενός πληροφοριακού συστήματος| και ανάπτυξη συστήματος)?$",				
102
				"(?i)^Consultants' contract(s)?$",
103
				"(?i)^Upute autorima$",
104
				"(?i)^Bijdrage tot de Kennis van den Godsdienst der Dajaks van Lan(d|f)ak en Tajan$",
105
				"^Joshi shin kokubun$",
106
				"^Kōtō shōgaku dokuhon nōson'yō$",
107
				"^Jinjō shōgaku shōka$",
108
				"^Shōgaku shūjichō$",
109
				"^Nihon joshi dokuhon$",
110
				"^Joshi shin dokuhon$",
111
				"^Chūtō kanbun dokuhon$",
112
				"^Wabun dokuhon$",
113
				"(?i)^(Analysis of economy selected village or town|Rozbor hospodaření vybrané obce či města)$",
114
				"(?i)^cardiac rehabilitation$",
115
				"(?i)^Analytical summary$",
116
				"^Thesaurus resolutionum Sacrae Congregationis Concilii$",
117
				"(?i)^Sumario analítico(\\s{1})?(Analitic summary)?$", 
118
				"^Prikazi i osvrti$",
119
				"^Rodinný dům s provozovnou$",
120
				"^Family house with an establishment$",
121
				"^Shinsei chūtō shin kokugun$",
122
				"^Pulmonary alveolar proteinosis(\\.?)$",
123
				"^Shinshū kanbun$",
124
				"^Viñeta(s?) de Rodríguez$",
125
				"(?i)^RUBRIKA UREDNIKA$",
126
				"^A Matching Model of the Academic Publication Market$",
127
				"^Yōgaku kōyō$",
128
			
129
				"^Internetový marketing$",
130
				"^Internet marketing$",
131
				"^Chūtō kokugo dokuhon$",
132
				"^Kokugo dokuhon$",
133
				"^Antibiotic Cover for Dental Extraction(s?)$",
134
				"^Strategie podniku$",				
135
				"^Strategy of an Enterprise$",
136
				"(?i)^respiratory disease(s?)(\.?)$",
137
				"^Award(s?) for Gallantry in Civil Defence$",
138
				"^Podniková kultura$",
139
				"^Corporate Culture$",
140
				"^Severe hyponatraemia in hospital inpatient(s?)(\.?)$",
141
				"^Pracovní motivace$",
142
				"^Work Motivation$",
143
				"^Kaitei kōtō jogaku dokuhon$",
144
				"^Konsolidovaná účetní závěrka$",
145
				"^Consolidated Financial Statements$",
146
				"(?i)^intracranial tumour(s?)$",
147
				"^Climate Change Mitigation Options and Directed Technical Change: A Decentralized Equilibrium Analysis$",
148
				"^\\[CERVECERIAS MAHOU(\\.|\\:) INTERIOR\\] \\[Material gráfico\\]$",
149
				"^Housing Market Dynamics(\\:|\\.) On the Contribution of Income Shocks and Credit Constraint(s?)$",
150
				"^\\[Funciones auxiliares de la música en Radio París,.*\\]$",
151
				"^Úroveň motivačního procesu jako způsobu vedení lidí$",
152
				"^The level of motivation process as a leadership$",
153
				"^Pay-beds in N(\\.?)H(\\.?)S(\\.?) Hospitals$",
154
				"(?i)^news and events$",
155
				"(?i)^NOVOSTI I DOGAĐAJI$",
156
				"^Sansū no gakushū$",
157
				"^Posouzení informačního systému firmy a návrh změn$",
158
				"^Information System Assessment and Proposal for ICT Modification$",
159
				"^Stresové zatížení pracovníků ve vybrané profesi$",
160
				"^Stress load in a specific job$",
161
				
162
				"^Sunday: Poster Sessions, Pt.*$",
163
				"^Monday: Poster Sessions, Pt.*$",
164
				"^Wednesday: Poster Sessions, Pt.*",
165
				"^Tuesday: Poster Sessions, Pt.*$",
166
				
167
				"^Analýza reklamy$",
168
				"^Analysis of advertising$",
169
			
170
				"^Shōgaku shūshinsho$",
171
				"^Shōgaku sansū$",
172
				"^Shintei joshi kokubun$",
173
				"^Taishō joshi kokubun dokuhon$",
174
				"^Joshi kokubun$",				
175
												
176
				"^Účetní uzávěrka a účetní závěrka v ČR$",
177
				"(?i)^The \"?Causes\"? of Cancer$",
178
				"^Normas para la publicación de artículos$",
179
				"^Editor('|s)(s|') [Rr]eply$",
180
				"^Editor(’|s)(s|’) letter$",
181
				"^Redaktoriaus žodis$",		
182
				"^DISCUSSION ON THE PRECEDING PAPER$",
183
				"^Kōtō shōgaku shūshinsho jidōyō$",
184
				"^Shōgaku nihon rekishi$",
185
				"^(Theory of the flow of action currents in isolated myelinated nerve fibers).*$",
186
				"^Préface$",
187
				"^Occupational [Hh]ealth [Ss]ervices.$",
188
				"^In Memoriam Professor Toshiyuki TAKESHIMA$",
189
				"^Účetní závěrka ve vybraném podniku.*$",
190
				"^Financial statements in selected company$",
191
				"^Abdominal [Aa]ortic [Aa]neurysms.*$",
192
				"^Pseudomyxoma peritonei$",
193
				"^Kazalo autora$",			
194
			
195
				"(?i)^uvodna riječ$",
196
				"^Motivace jako způsob vedení lidí$",
197
				"^Motivation as a leadership$",
198
				"^Polyfunkční dům$",
199
				"^Multi\\-funkcional building$",
200
				"^Podnikatelský plán$",
201
				"^Business Plan$",
202
				"^Oceňování nemovitostí$",
203
				"^Marketingová komunikace$",
204
				"^Marketing communication$",
205
				"^Sumario Analítico$",
206
				"^Riječ uredništva$",
207
				"^Savjetovanja i priredbe$",
208
				"^Índice$",
209
				"^(Starobosanski nadpisi).*$",
210
				"^Vzdělávání pracovníků v organizaci$",
211
				"^Staff training in organization$",
212
				"^(Life Histories of North American Geometridae).*$",
213
				"^Strategická analýza podniku$",
214
				"^Strategic Analysis of an Enterprise$",
215
				"^Sadržaj$",
216
				"^Upute suradnicima$",
217
				"^Rodinný dům$",
218
				"(?i)^Fami(l)?ly house$",
219
				"^Upute autorima$",
220
				"^Strategic Analysis$",
221
				"^Finanční analýza vybraného podniku$",
222
				"^Finanční analýza$",
223
				"^Riječ urednika$",
224
				"(?i)^Content(s?)$",
225
				"(?i)^Inhalt$",
226
				"^Jinjō shōgaku shūshinsho jidōyō$",
227
				"(?i)^Index$",
228
				"^Chūgaku kokubun kyōkasho$",
229
				"^Retrato de una mujer$",
230
				"^Retrato de un hombre$",
231
				"^Kōtō shōgaku dokuhon$",
232
				"^Shotōka kokugo$",
233
				"^Shōgaku dokuhon$",
234
				"^Jinjō shōgaku kokugo dokuhon$",
235
				"^Shinsei kokugo dokuhon$",
236
				"^Teikoku dokuhon$",
237
				"^Instructions to Authors$",
238
				"^KİTAP TAHLİLİ$",
239
				"^PRZEGLĄD PIŚMIENNICTWA$",
240
				"(?i)^Presentación$",
241
				"^İçindekiler$",
242
				"(?i)^Tabl?e of contents$",
243
				"^(CODICE DEL BEATO DE LOS REYES FERNANDO I Y SANCHA).*$",
244
				"^(\\[MADRID\\. BIBL\\. NAC\\. N.*KING FERDINAND I.*FROM SAN ISIDORO DE LEON\\. FACUNDUS SCRIPSIT DATED.*\\]).*",
245
				"^Editorial( Board)?$",
246
				"(?i)^Editorial \\(English\\)$",
247
				"^Editörden$",			
248
				"^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
249
				"^(Kiri Karl Morgensternile).*$",
250
				"^(\\[Eksliibris Aleksandr).*\\]$",
251
				"^(\\[Eksliibris Aleksandr).*$",
252
				"^(Eksliibris Aleksandr).*$",
253
				"^(Kiri A\\. de Vignolles).*$",
254
				"^(2 kirja Karl Morgensternile).*$",
255
				"^(Pirita kloostri idaosa arheoloogilised).*$",
256
				"^(Kiri tundmatule).*$",
257
				"^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
258
				"^(Eksliibris Nikolai Birukovile).*$",
259
				"^(Eksliibris Nikolai Issakovile).*$",
260
				"^(WHP Cruise Summary Information of section).*$",
261
				"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
262
				"^(Measurement of the spin\\-dependent structure function).*"
263
			]
264
		} 		
265
	}
266
}
(5-5/6)