|
1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_primary_processing">
|
|
2
|
|
|
3
|
<parameters>
|
|
4
|
<property>
|
|
5
|
<name>remove_sideproducts</name>
|
|
6
|
<value>true</value>
|
|
7
|
<description>flag indicating inference side products will be erased</description>
|
|
8
|
</property>
|
|
9
|
<!-- processing modes -->
|
|
10
|
<property>
|
|
11
|
<name>active_referenceextraction_project</name>
|
|
12
|
<value>true</value>
|
|
13
|
<description>flag indicating project reference extraction should be enabled</description>
|
|
14
|
</property>
|
|
15
|
<property>
|
|
16
|
<name>active_referenceextraction_dataset</name>
|
|
17
|
<value>true</value>
|
|
18
|
<description>flag indicating dataset reference extraction should be enabled</description>
|
|
19
|
</property>
|
|
20
|
<property>
|
|
21
|
<!-- currently disabled by default -->
|
|
22
|
<name>active_referenceextraction_researchinitiative</name>
|
|
23
|
<value>false</value>
|
|
24
|
<description>flag indicating researchinitiative reference extraction should be enabled</description>
|
|
25
|
</property>
|
|
26
|
<property>
|
|
27
|
<!-- currently disabled by default -->
|
|
28
|
<name>active_referenceextraction_pdb</name>
|
|
29
|
<value>false</value>
|
|
30
|
<description>flag indicating protein databank reference extraction should be enabled</description>
|
|
31
|
</property>
|
|
32
|
<property>
|
|
33
|
<name>active_documentsclassification</name>
|
|
34
|
<value>true</value>
|
|
35
|
<description>flag indicating documents classification should be enabled</description>
|
|
36
|
</property>
|
|
37
|
<property>
|
|
38
|
<name>active_documentssimilarity</name>
|
|
39
|
<value>true</value>
|
|
40
|
<description>flag indicating documents similarity should be enabled</description>
|
|
41
|
</property>
|
|
42
|
<property>
|
|
43
|
<name>active_citationmatching</name>
|
|
44
|
<!-- currently disabled by default -->
|
|
45
|
<value>false</value>
|
|
46
|
<description>flag indicating citation matching should be enabled</description>
|
|
47
|
</property>
|
|
48
|
<property>
|
|
49
|
<name>active_statistics</name>
|
|
50
|
<!-- currently disabled by default -->
|
|
51
|
<value>false</value>
|
|
52
|
<description>flag indicating statistics generation should be enabled</description>
|
|
53
|
</property>
|
|
54
|
<!-- input ports -->
|
|
55
|
<property>
|
|
56
|
<name>input_document_metadata</name>
|
|
57
|
<description>input document metadata directory</description>
|
|
58
|
</property>
|
|
59
|
<property>
|
|
60
|
<name>input_document_to_project</name>
|
|
61
|
<description>input document to project relation directory</description>
|
|
62
|
</property>
|
|
63
|
<property>
|
|
64
|
<name>input_document_text</name>
|
|
65
|
<description>input document text directory</description>
|
|
66
|
</property>
|
|
67
|
<property>
|
|
68
|
<name>input_document_text_wos</name>
|
|
69
|
<description>input document text directory holding WOS records</description>
|
|
70
|
</property>
|
|
71
|
<property>
|
|
72
|
<name>input_project</name>
|
|
73
|
<description>input project directory</description>
|
|
74
|
</property>
|
|
75
|
<property>
|
|
76
|
<name>input_person</name>
|
|
77
|
<description>input person directory</description>
|
|
78
|
</property>
|
|
79
|
<property>
|
|
80
|
<name>input_dataset</name>
|
|
81
|
<description>input dataset directory</description>
|
|
82
|
</property>
|
|
83
|
<property>
|
|
84
|
<name>input_extracted_document_metadata</name>
|
|
85
|
<description>input extracted document metadata directory</description>
|
|
86
|
</property>
|
|
87
|
<property>
|
|
88
|
<name>input_citation_pmc</name>
|
|
89
|
<description>input directory holding citations extracted from PMC</description>
|
|
90
|
</property>
|
|
91
|
<property>
|
|
92
|
<name>input_project_concept</name>
|
|
93
|
<description>input project concept directory</description>
|
|
94
|
</property>
|
|
95
|
<!-- citation matching related -->
|
|
96
|
<property>
|
|
97
|
<name>cit_genAuthorIdxJavaOpts</name>
|
|
98
|
<value>-Xmx8g</value>
|
|
99
|
<description>java opts for author index creation for citation purposes</description>
|
|
100
|
</property>
|
|
101
|
<!-- document similarity related -->
|
|
102
|
<property>
|
|
103
|
<name>ds_parallel</name>
|
|
104
|
<value>20</value>
|
|
105
|
<description>document similarity pig parallel</description>
|
|
106
|
</property>
|
|
107
|
<property>
|
|
108
|
<name>ds_mapredChildJavaOpts</name>
|
|
109
|
<value>-Xmx20g</value>
|
|
110
|
<description>mapred child java opts</description>
|
|
111
|
</property>
|
|
112
|
<property>
|
|
113
|
<name>ds_sample</name>
|
|
114
|
<value>1.0</value>
|
|
115
|
<description>sample rate</description>
|
|
116
|
</property>
|
|
117
|
<property>
|
|
118
|
<name>ds_removal_rate</name>
|
|
119
|
<value>0.99</value>
|
|
120
|
<description>document similarity removal rate</description>
|
|
121
|
</property>
|
|
122
|
<property>
|
|
123
|
<name>ds_removal_least_used</name>
|
|
124
|
<value>20</value>
|
|
125
|
<description>document similarity least used removal</description>
|
|
126
|
</property>
|
|
127
|
<property>
|
|
128
|
<name>ds_tfidfTopnTermPerDocument</name>
|
|
129
|
<value>20</value>
|
|
130
|
</property>
|
|
131
|
<property>
|
|
132
|
<name>ds_similarityTopnDocumentPerDocument</name>
|
|
133
|
<value>20</value>
|
|
134
|
</property>
|
|
135
|
<!-- output ports -->
|
|
136
|
<property>
|
|
137
|
<name>output_document_to_project</name>
|
|
138
|
<description>project reference extraction output directory</description>
|
|
139
|
</property>
|
|
140
|
<property>
|
|
141
|
<name>output_document_to_project_concepts</name>
|
|
142
|
<description>document to project concepts output directory</description>
|
|
143
|
</property>
|
|
144
|
<property>
|
|
145
|
<name>output_document_to_dataset</name>
|
|
146
|
<description>dataset reference extraction output directory</description>
|
|
147
|
</property>
|
|
148
|
<property>
|
|
149
|
<name>output_document_to_research_initiatives</name>
|
|
150
|
<description>research initiatives reference extraction output directory</description>
|
|
151
|
</property>
|
|
152
|
<property>
|
|
153
|
<name>output_document_to_pdb</name>
|
|
154
|
<description>protein databank reference extraction output directory</description>
|
|
155
|
</property>
|
|
156
|
<property>
|
|
157
|
<name>output_document_to_document_classes</name>
|
|
158
|
<description>output document classification directory</description>
|
|
159
|
</property>
|
|
160
|
<property>
|
|
161
|
<name>output_citation</name>
|
|
162
|
<description>output containing grouped citations coming from citation matching and pmc ingestion</description>
|
|
163
|
</property>
|
|
164
|
<property>
|
|
165
|
<name>output_document_similarity</name>
|
|
166
|
<description>output document similarity directory</description>
|
|
167
|
</property>
|
|
168
|
<property>
|
|
169
|
<name>output_document_statistics</name>
|
|
170
|
<description>output document statistics directory</description>
|
|
171
|
</property>
|
|
172
|
<property>
|
|
173
|
<name>output_author_statistics</name>
|
|
174
|
<description>output author statistics directory</description>
|
|
175
|
</property>
|
|
176
|
<property>
|
|
177
|
<name>output_project_statistics</name>
|
|
178
|
<description>output project statistics directory</description>
|
|
179
|
</property>
|
|
180
|
</parameters>
|
|
181
|
|
|
182
|
<global>
|
|
183
|
<job-tracker>${jobTracker}</job-tracker>
|
|
184
|
<name-node>${nameNode}</name-node>
|
|
185
|
<configuration>
|
|
186
|
<property>
|
|
187
|
<name>mapred.job.queue.name</name>
|
|
188
|
<value>${queueName}</value>
|
|
189
|
</property>
|
|
190
|
</configuration>
|
|
191
|
</global>
|
|
192
|
|
|
193
|
<start to="forking" />
|
|
194
|
|
|
195
|
<fork name="forking">
|
|
196
|
<path start="decision-referenceextraction_project"/>
|
|
197
|
<path start="decision-referenceextraction_dataset"/>
|
|
198
|
<path start="decision-referenceextraction_researchinitiative"/>
|
|
199
|
<path start="decision-referenceextraction_pdb"/>
|
|
200
|
<path start="transformers_metadatamerger"/>
|
|
201
|
</fork>
|
|
202
|
|
|
203
|
<!-- start of project reference extraction block -->
|
|
204
|
<decision name="decision-referenceextraction_project">
|
|
205
|
<switch>
|
|
206
|
<case to="referenceextraction_project">${active_referenceextraction_project eq "true"}</case>
|
|
207
|
<default to="skip-referenceextraction_project"/>
|
|
208
|
</switch>
|
|
209
|
</decision>
|
|
210
|
|
|
211
|
<action name="referenceextraction_project">
|
|
212
|
<sub-workflow>
|
|
213
|
<app-path>${wf:appPath()}/referenceextraction_project</app-path>
|
|
214
|
<propagate-configuration/>
|
|
215
|
<configuration>
|
|
216
|
<property>
|
|
217
|
<name>workingDir</name>
|
|
218
|
<value>${workingDir}/referenceextraction_project/working_dir</value>
|
|
219
|
</property>
|
|
220
|
<property>
|
|
221
|
<name>input_document_text</name>
|
|
222
|
<value>${input_document_text}</value>
|
|
223
|
</property>
|
|
224
|
<property>
|
|
225
|
<name>input_project</name>
|
|
226
|
<value>${input_project}</value>
|
|
227
|
</property>
|
|
228
|
<property>
|
|
229
|
<name>output_document_to_project</name>
|
|
230
|
<!-- referenceextraction_project directory is created at subworkflow prepare phase -->
|
|
231
|
<value>${output_document_to_project}</value>
|
|
232
|
</property>
|
|
233
|
</configuration>
|
|
234
|
</sub-workflow>
|
|
235
|
<ok to="transformers_project_toconcept"/>
|
|
236
|
<error to="fail" />
|
|
237
|
</action>
|
|
238
|
|
|
239
|
<action name="transformers_project_toconcept">
|
|
240
|
<sub-workflow>
|
|
241
|
<app-path>${wf:appPath()}/transformers_project_toconcept</app-path>
|
|
242
|
<propagate-configuration/>
|
|
243
|
<configuration>
|
|
244
|
<property>
|
|
245
|
<name>workingDir</name>
|
|
246
|
<value>${workingDir}/transformers_project_toconcept/working_dir</value>
|
|
247
|
</property>
|
|
248
|
<property>
|
|
249
|
<name>input_document_to_project</name>
|
|
250
|
<value>${output_document_to_project}</value>
|
|
251
|
</property>
|
|
252
|
<property>
|
|
253
|
<name>input_project</name>
|
|
254
|
<value>${input_project}</value>
|
|
255
|
</property>
|
|
256
|
<property>
|
|
257
|
<name>input_concept</name>
|
|
258
|
<value>${input_project_concept}</value>
|
|
259
|
</property>
|
|
260
|
<property>
|
|
261
|
<name>output</name>
|
|
262
|
<value>${workingDir}/transformers_project_toconcept/out</value>
|
|
263
|
</property>
|
|
264
|
</configuration>
|
|
265
|
</sub-workflow>
|
|
266
|
<ok to="transformers_concept_to_researchinitiatives"/>
|
|
267
|
<error to="fail" />
|
|
268
|
</action>
|
|
269
|
|
|
270
|
<action name="transformers_concept_to_researchinitiatives">
|
|
271
|
<sub-workflow>
|
|
272
|
<app-path>${wf:appPath()}/transformers_export_researchinitiatives</app-path>
|
|
273
|
<propagate-configuration/>
|
|
274
|
<configuration>
|
|
275
|
<property>
|
|
276
|
<name>workingDir</name>
|
|
277
|
<value>${workingDir}/transformers_concept_to_researchinitiatives/working_dir</value>
|
|
278
|
</property>
|
|
279
|
<property>
|
|
280
|
<name>input_document_to_research_initiative</name>
|
|
281
|
<value>${workingDir}/transformers_project_toconcept/out</value>
|
|
282
|
</property>
|
|
283
|
<property>
|
|
284
|
<name>output_document_to_research_initiatives</name>
|
|
285
|
<value>${output_document_to_project_concepts}</value>
|
|
286
|
</property>
|
|
287
|
</configuration>
|
|
288
|
</sub-workflow>
|
|
289
|
<ok to="joining"/>
|
|
290
|
<error to="fail" />
|
|
291
|
</action>
|
|
292
|
|
|
293
|
<action name="skip-referenceextraction_project">
|
|
294
|
<java>
|
|
295
|
<prepare>
|
|
296
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
297
|
<delete path="${nameNode}${workingDir}/referenceextraction_project" />
|
|
298
|
<delete path="${nameNode}${workingDir}/transformers_project_toconcept" />
|
|
299
|
<delete path="${nameNode}${workingDir}/transformers_concept_to_researchinitiatives" />
|
|
300
|
<delete path="${nameNode}${output_document_to_project}"/>
|
|
301
|
<delete path="${nameNode}${output_document_to_project_concepts}"/>
|
|
302
|
<mkdir path="${nameNode}${workingDir}/referenceextraction_project" />
|
|
303
|
<mkdir path="${nameNode}${output_document_to_project}"/>
|
|
304
|
<mkdir path="${nameNode}${output_document_to_project_concepts}"/>
|
|
305
|
</prepare>
|
|
306
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
307
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
308
|
<arg>-C{referenceextraction_project,
|
|
309
|
eu.dnetlib.iis.referenceextraction.project.schemas.DocumentToProject,
|
|
310
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
311
|
<arg>-C{document_to_project_concepts,
|
|
312
|
eu.dnetlib.iis.export.schemas.DocumentToConceptIds,
|
|
313
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
314
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
315
|
<arg>-Oreferenceextraction_project=${output_document_to_project}</arg>
|
|
316
|
<arg>-Odocument_to_project_concepts=${output_document_to_project_concepts}</arg>
|
|
317
|
</java>
|
|
318
|
<ok to="joining"/>
|
|
319
|
<error to="fail"/>
|
|
320
|
</action>
|
|
321
|
<!-- end of project reference extraction block -->
|
|
322
|
|
|
323
|
<!-- start of dataset reference extraction block -->
|
|
324
|
<decision name="decision-referenceextraction_dataset">
|
|
325
|
<switch>
|
|
326
|
<case to="referenceextraction_dataset">${active_referenceextraction_dataset eq "true"}</case>
|
|
327
|
<default to="skip-referenceextraction_dataset"/>
|
|
328
|
</switch>
|
|
329
|
</decision>
|
|
330
|
|
|
331
|
<action name="referenceextraction_dataset">
|
|
332
|
<sub-workflow>
|
|
333
|
<app-path>${wf:appPath()}/referenceextraction_dataset</app-path>
|
|
334
|
<propagate-configuration/>
|
|
335
|
<configuration>
|
|
336
|
<property>
|
|
337
|
<name>workingDir</name>
|
|
338
|
<value>${workingDir}/referenceextraction_dataset/working_dir</value>
|
|
339
|
</property>
|
|
340
|
<property>
|
|
341
|
<name>input_document_text</name>
|
|
342
|
<value>${input_document_text}</value>
|
|
343
|
</property>
|
|
344
|
<property>
|
|
345
|
<name>input_dataset</name>
|
|
346
|
<value>${input_dataset}</value>
|
|
347
|
</property>
|
|
348
|
<property>
|
|
349
|
<name>output_document_to_dataset</name>
|
|
350
|
<!-- referenceextraction_dataset directory is created at subworkflow prepare phase -->
|
|
351
|
<value>${output_document_to_dataset}</value>
|
|
352
|
</property>
|
|
353
|
</configuration>
|
|
354
|
</sub-workflow>
|
|
355
|
<ok to="joining"/>
|
|
356
|
<error to="fail" />
|
|
357
|
</action>
|
|
358
|
|
|
359
|
<action name="skip-referenceextraction_dataset">
|
|
360
|
<java>
|
|
361
|
<prepare>
|
|
362
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
363
|
<delete path="${nameNode}${workingDir}/referenceextraction_dataset" />
|
|
364
|
<delete path="${nameNode}${output_document_to_dataset}"/>
|
|
365
|
<mkdir path="${nameNode}${workingDir}/referenceextraction_dataset" />
|
|
366
|
<mkdir path="${nameNode}${output_document_to_dataset}"/>
|
|
367
|
</prepare>
|
|
368
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
369
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
370
|
<arg>-C{referenceextraction_dataset,
|
|
371
|
eu.dnetlib.iis.referenceextraction.dataset.schemas.DocumentToDataSet,
|
|
372
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
373
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
374
|
<arg>-Oreferenceextraction_dataset=${output_document_to_dataset}</arg>
|
|
375
|
</java>
|
|
376
|
<ok to="joining"/>
|
|
377
|
<error to="fail"/>
|
|
378
|
</action>
|
|
379
|
<!-- end of dataset reference extraction block -->
|
|
380
|
|
|
381
|
<!-- start of researchinitiative reference extraction block -->
|
|
382
|
<decision name="decision-referenceextraction_researchinitiative">
|
|
383
|
<switch>
|
|
384
|
<case to="referenceextraction_researchinitiative_collapser">${active_referenceextraction_researchinitiative eq "true"}</case>
|
|
385
|
<default to="skip-referenceextraction_researchinitiative"/>
|
|
386
|
</switch>
|
|
387
|
</decision>
|
|
388
|
|
|
389
|
<action name="referenceextraction_researchinitiative_collapser">
|
|
390
|
<sub-workflow>
|
|
391
|
<app-path>${wf:appPath()}/collapsers_multiple_input_collapser</app-path>
|
|
392
|
<propagate-configuration/>
|
|
393
|
<configuration>
|
|
394
|
<property>
|
|
395
|
<name>workingDir</name>
|
|
396
|
<value>${workingDir}/referenceextraction_researchinitiative_collapser/working_dir</value>
|
|
397
|
</property>
|
|
398
|
<property>
|
|
399
|
<name>origin_1</name>
|
|
400
|
<value>document_text</value>
|
|
401
|
</property>
|
|
402
|
<property>
|
|
403
|
<name>input_1</name>
|
|
404
|
<value>${input_document_text}</value>
|
|
405
|
</property>
|
|
406
|
<property>
|
|
407
|
<name>origin_2</name>
|
|
408
|
<value>document_text_wos</value>
|
|
409
|
</property>
|
|
410
|
<property>
|
|
411
|
<name>input_2</name>
|
|
412
|
<value>${input_document_text_wos}</value>
|
|
413
|
</property>
|
|
414
|
<property>
|
|
415
|
<name>blocking_field</name>
|
|
416
|
<value>id</value>
|
|
417
|
</property>
|
|
418
|
<property>
|
|
419
|
<name>schema_input</name>
|
|
420
|
<value>eu.dnetlib.iis.metadataextraction.schemas.DocumentText</value>
|
|
421
|
</property>
|
|
422
|
<property>
|
|
423
|
<name>output</name>
|
|
424
|
<value>${workingDir}/referenceextraction_researchinitiative_collapser/output</value>
|
|
425
|
</property>
|
|
426
|
<property>
|
|
427
|
<name>schema_input_envelope</name>
|
|
428
|
<value>eu.dnetlib.iis.collapsers.schemas.DocumentTextEnvelope</value>
|
|
429
|
</property>
|
|
430
|
<property>
|
|
431
|
<name>record_collapser</name>
|
|
432
|
<value>eu.dnetlib.iis.collapsers.origins.DocumentTextCollapser</value>
|
|
433
|
</property>
|
|
434
|
</configuration>
|
|
435
|
</sub-workflow>
|
|
436
|
<ok to="referenceextraction_researchinitiative"/>
|
|
437
|
<error to="fail"/>
|
|
438
|
</action>
|
|
439
|
|
|
440
|
<action name="referenceextraction_researchinitiative">
|
|
441
|
<sub-workflow>
|
|
442
|
<app-path>${wf:appPath()}/referenceextraction_researchinitiative</app-path>
|
|
443
|
<propagate-configuration/>
|
|
444
|
<configuration>
|
|
445
|
<property>
|
|
446
|
<name>workingDir</name>
|
|
447
|
<value>${workingDir}/referenceextraction_researchinitiative/working_dir</value>
|
|
448
|
</property>
|
|
449
|
<property>
|
|
450
|
<name>input_document_text</name>
|
|
451
|
<value>${workingDir}/referenceextraction_researchinitiative_collapser/output</value>
|
|
452
|
</property>
|
|
453
|
<property>
|
|
454
|
<name>output_document_to_research_initiative</name>
|
|
455
|
<value>${workingDir}/referenceextraction_researchinitiative/output</value>
|
|
456
|
</property>
|
|
457
|
</configuration>
|
|
458
|
</sub-workflow>
|
|
459
|
<ok to="transformers_export_researchinitiatives"/>
|
|
460
|
<error to="fail" />
|
|
461
|
</action>
|
|
462
|
|
|
463
|
<action name="transformers_export_researchinitiatives">
|
|
464
|
<sub-workflow>
|
|
465
|
<app-path>${wf:appPath()}/transformers_export_researchinitiatives</app-path>
|
|
466
|
<propagate-configuration/>
|
|
467
|
<configuration>
|
|
468
|
<property>
|
|
469
|
<name>workingDir</name>
|
|
470
|
<value>${workingDir}/transformers_export_researchinitiatives/working_dir</value>
|
|
471
|
</property>
|
|
472
|
<property>
|
|
473
|
<name>input_document_to_research_initiative</name>
|
|
474
|
<value>${workingDir}/referenceextraction_researchinitiative/output</value>
|
|
475
|
</property>
|
|
476
|
<property>
|
|
477
|
<name>output_document_to_research_initiatives</name>
|
|
478
|
<value>${output_document_to_research_initiatives}</value>
|
|
479
|
</property>
|
|
480
|
</configuration>
|
|
481
|
</sub-workflow>
|
|
482
|
|
|
483
|
<ok to="joining"/>
|
|
484
|
<error to="fail" />
|
|
485
|
</action>
|
|
486
|
|
|
487
|
<action name="skip-referenceextraction_researchinitiative">
|
|
488
|
<java>
|
|
489
|
<prepare>
|
|
490
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
491
|
<delete path="${nameNode}${workingDir}/transformers_export_researchinitiatives" />
|
|
492
|
<delete path="${nameNode}${output_document_to_research_initiatives}"/>
|
|
493
|
<mkdir path="${nameNode}${workingDir}/transformers_export_researchinitiatives" />
|
|
494
|
<mkdir path="${nameNode}${output_document_to_research_initiatives}"/>
|
|
495
|
</prepare>
|
|
496
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
497
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
498
|
<arg>-C{referenceextraction_researchinitiatives,
|
|
499
|
eu.dnetlib.iis.export.schemas.DocumentToConceptIds,
|
|
500
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
501
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
502
|
<arg>-Oreferenceextraction_researchinitiatives=${output_document_to_research_initiatives}</arg>
|
|
503
|
</java>
|
|
504
|
<ok to="joining"/>
|
|
505
|
<error to="fail"/>
|
|
506
|
</action>
|
|
507
|
<!-- end of researchinitiative reference extraction block -->
|
|
508
|
|
|
509
|
<!-- start of pdb reference extraction block -->
|
|
510
|
<decision name="decision-referenceextraction_pdb">
|
|
511
|
<switch>
|
|
512
|
<case to="referenceextraction_pdb">${active_referenceextraction_pdb eq "true"}</case>
|
|
513
|
<default to="skip-referenceextraction_pdb"/>
|
|
514
|
</switch>
|
|
515
|
</decision>
|
|
516
|
|
|
517
|
<action name="referenceextraction_pdb">
|
|
518
|
<sub-workflow>
|
|
519
|
<app-path>${wf:appPath()}/referenceextraction_pdb</app-path>
|
|
520
|
<propagate-configuration/>
|
|
521
|
<configuration>
|
|
522
|
<property>
|
|
523
|
<name>workingDir</name>
|
|
524
|
<value>${workingDir}/referenceextraction_pdb/working_dir</value>
|
|
525
|
</property>
|
|
526
|
<property>
|
|
527
|
<name>input_document_text</name>
|
|
528
|
<value>${input_document_text}</value>
|
|
529
|
</property>
|
|
530
|
<property>
|
|
531
|
<name>output</name>
|
|
532
|
<value>${workingDir}/referenceextraction_pdb/output</value>
|
|
533
|
</property>
|
|
534
|
</configuration>
|
|
535
|
</sub-workflow>
|
|
536
|
<ok to="transformers_export_pdb"/>
|
|
537
|
<error to="fail" />
|
|
538
|
</action>
|
|
539
|
|
|
540
|
<action name="transformers_export_pdb">
|
|
541
|
<sub-workflow>
|
|
542
|
<app-path>${wf:appPath()}/transformers_export_researchinitiatives</app-path>
|
|
543
|
<propagate-configuration/>
|
|
544
|
<configuration>
|
|
545
|
<property>
|
|
546
|
<name>workingDir</name>
|
|
547
|
<value>${workingDir}/transformers_export_pdb/working_dir</value>
|
|
548
|
</property>
|
|
549
|
<property>
|
|
550
|
<name>input_document_to_research_initiative</name>
|
|
551
|
<value>${workingDir}/referenceextraction_pdb/output</value>
|
|
552
|
</property>
|
|
553
|
<property>
|
|
554
|
<name>output_document_to_research_initiatives</name>
|
|
555
|
<value>${output_document_to_pdb}</value>
|
|
556
|
</property>
|
|
557
|
</configuration>
|
|
558
|
</sub-workflow>
|
|
559
|
|
|
560
|
<ok to="joining"/>
|
|
561
|
<error to="fail" />
|
|
562
|
</action>
|
|
563
|
|
|
564
|
<action name="skip-referenceextraction_pdb">
|
|
565
|
<java>
|
|
566
|
<prepare>
|
|
567
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
568
|
<delete path="${nameNode}${workingDir}/transformers_export_pdb" />
|
|
569
|
<delete path="${nameNode}${output_document_to_pdb}"/>
|
|
570
|
<mkdir path="${nameNode}${workingDir}/transformers_export_pdb" />
|
|
571
|
<mkdir path="${nameNode}${output_document_to_pdb}"/>
|
|
572
|
</prepare>
|
|
573
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
574
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
575
|
<arg>-C{referenceextraction_pdb,
|
|
576
|
eu.dnetlib.iis.export.schemas.DocumentToConceptIds,
|
|
577
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
578
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
579
|
<arg>-Oreferenceextraction_pdb=${output_document_to_pdb}</arg>
|
|
580
|
</java>
|
|
581
|
<ok to="joining"/>
|
|
582
|
<error to="fail"/>
|
|
583
|
</action>
|
|
584
|
<!-- end of pdb reference extraction block -->
|
|
585
|
|
|
586
|
<!-- metadatamerger branch -->
|
|
587
|
<action name="transformers_metadatamerger">
|
|
588
|
<sub-workflow>
|
|
589
|
<app-path>${wf:appPath()}/transformers_metadatamerger</app-path>
|
|
590
|
<propagate-configuration/>
|
|
591
|
<configuration>
|
|
592
|
<property>
|
|
593
|
<name>workingDir</name>
|
|
594
|
<value>${workingDir}/transformers_metadatamerger/working_dir</value>
|
|
595
|
</property>
|
|
596
|
<property>
|
|
597
|
<name>input_base_metadata</name>
|
|
598
|
<value>${input_document_metadata}</value>
|
|
599
|
</property>
|
|
600
|
<property>
|
|
601
|
<name>input_extracted_metadata</name>
|
|
602
|
<value>${input_extracted_document_metadata}</value>
|
|
603
|
</property>
|
|
604
|
<property>
|
|
605
|
<name>output_merged_metadata</name>
|
|
606
|
<value>${workingDir}/transformers_metadatamerger/output_merged_metadata</value>
|
|
607
|
</property>
|
|
608
|
</configuration>
|
|
609
|
</sub-workflow>
|
|
610
|
|
|
611
|
<ok to="decision-documentsclassification"/>
|
|
612
|
<error to="fail" />
|
|
613
|
</action>
|
|
614
|
|
|
615
|
<!-- start of documents classification part -->
|
|
616
|
<decision name="decision-documentsclassification">
|
|
617
|
<switch>
|
|
618
|
<case to="transformers_documentsclassification">${active_documentsclassification eq "true"}</case>
|
|
619
|
<default to="skip-documentsclassification"/>
|
|
620
|
</switch>
|
|
621
|
</decision>
|
|
622
|
|
|
623
|
<action name="transformers_documentsclassification">
|
|
624
|
<sub-workflow>
|
|
625
|
<app-path>${wf:appPath()}/transformers_documentsclassification</app-path>
|
|
626
|
<propagate-configuration/>
|
|
627
|
<configuration>
|
|
628
|
<property>
|
|
629
|
<name>workingDir</name>
|
|
630
|
<value>${workingDir}/transformers_documentsclassification/working_dir</value>
|
|
631
|
</property>
|
|
632
|
<property>
|
|
633
|
<name>input_merged_metadata</name>
|
|
634
|
<value>${workingDir}/transformers_metadatamerger/output_merged_metadata</value>
|
|
635
|
</property>
|
|
636
|
<property>
|
|
637
|
<name>output_document_metadata</name>
|
|
638
|
<value>${workingDir}/transformers_documentsclassification/output_document_metadata</value>
|
|
639
|
</property>
|
|
640
|
</configuration>
|
|
641
|
</sub-workflow>
|
|
642
|
<ok to="documentsclassification_main"/>
|
|
643
|
<error to="fail" />
|
|
644
|
</action>
|
|
645
|
|
|
646
|
<action name="documentsclassification_main">
|
|
647
|
<sub-workflow>
|
|
648
|
<app-path>${wf:appPath()}/documentsclassification_main</app-path>
|
|
649
|
<propagate-configuration/>
|
|
650
|
<configuration>
|
|
651
|
<property>
|
|
652
|
<name>workingDir</name>
|
|
653
|
<value>${workingDir}/documentsclassification_main/working_dir</value>
|
|
654
|
</property>
|
|
655
|
<property>
|
|
656
|
<name>input_document_metadata</name>
|
|
657
|
<value>${workingDir}/transformers_documentsclassification/output_document_metadata</value>
|
|
658
|
</property>
|
|
659
|
<property>
|
|
660
|
<name>output_document_to_document_classes</name>
|
|
661
|
<value>${output_document_to_document_classes}</value>
|
|
662
|
</property>
|
|
663
|
</configuration>
|
|
664
|
</sub-workflow>
|
|
665
|
<ok to="joining"/>
|
|
666
|
<error to="fail" />
|
|
667
|
</action>
|
|
668
|
|
|
669
|
<action name="skip-documentsclassification">
|
|
670
|
<java>
|
|
671
|
<prepare>
|
|
672
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
673
|
<delete path="${nameNode}${workingDir}/documentsclassification_main" />
|
|
674
|
<delete path="${nameNode}${output_document_to_document_classes}"/>
|
|
675
|
<mkdir path="${nameNode}${workingDir}/documentsclassification_main" />
|
|
676
|
<mkdir path="${nameNode}${output_document_to_document_classes}"/>
|
|
677
|
</prepare>
|
|
678
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
679
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
680
|
<arg>-C{documentsclassification,
|
|
681
|
eu.dnetlib.iis.documentsclassification.schemas.DocumentToDocumentClasses,
|
|
682
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
683
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
684
|
<arg>-Odocumentsclassification=${output_document_to_document_classes}</arg>
|
|
685
|
</java>
|
|
686
|
<ok to="joining"/>
|
|
687
|
<error to="fail"/>
|
|
688
|
</action>
|
|
689
|
<!-- end of documents classification part -->
|
|
690
|
|
|
691
|
<join name="joining" to="decision-citationmatching"/>
|
|
692
|
|
|
693
|
<!-- citation matching part -->
|
|
694
|
<decision name="decision-citationmatching">
|
|
695
|
<switch>
|
|
696
|
<case to="transformers_citationmatching">${active_citationmatching eq "true"}</case>
|
|
697
|
<default to="skip-citationmatching"/>
|
|
698
|
</switch>
|
|
699
|
</decision>
|
|
700
|
|
|
701
|
<action name="transformers_citationmatching">
|
|
702
|
<sub-workflow>
|
|
703
|
<app-path>${wf:appPath()}/transformers_citationmatching</app-path>
|
|
704
|
<propagate-configuration/>
|
|
705
|
<configuration>
|
|
706
|
<property>
|
|
707
|
<name>workingDir</name>
|
|
708
|
<value>${workingDir}/transformers_citationmatching/working_dir</value>
|
|
709
|
</property>
|
|
710
|
<property>
|
|
711
|
<name>input_metadata</name>
|
|
712
|
<value>${workingDir}/transformers_metadatamerger/output_merged_metadata</value>
|
|
713
|
</property>
|
|
714
|
<property>
|
|
715
|
<name>input_person</name>
|
|
716
|
<value>${input_person}</value>
|
|
717
|
</property>
|
|
718
|
<property>
|
|
719
|
<name>output_citation_metadata</name>
|
|
720
|
<value>${workingDir}/transformers_citationmatching/output_citation_metadata</value>
|
|
721
|
</property>
|
|
722
|
</configuration>
|
|
723
|
</sub-workflow>
|
|
724
|
<ok to="citationmatching_chain"/>
|
|
725
|
<error to="fail" />
|
|
726
|
</action>
|
|
727
|
|
|
728
|
<action name="citationmatching_chain">
|
|
729
|
<sub-workflow>
|
|
730
|
<app-path>${wf:appPath()}/citationmatching_chain</app-path>
|
|
731
|
<propagate-configuration/>
|
|
732
|
<configuration>
|
|
733
|
<property>
|
|
734
|
<name>workingDir</name>
|
|
735
|
<value>${workingDir}/citationmatching_chain/working_dir</value>
|
|
736
|
</property>
|
|
737
|
<property>
|
|
738
|
<name>input</name>
|
|
739
|
<value>${workingDir}/transformers_citationmatching/output_citation_metadata</value>
|
|
740
|
</property>
|
|
741
|
<property>
|
|
742
|
<name>output</name>
|
|
743
|
<value>${workingDir}/citationmatching_chain/output</value>
|
|
744
|
</property>
|
|
745
|
<property>
|
|
746
|
<name>cit_genAuthorIdxJavaOpts</name>
|
|
747
|
<value>${cit_genAuthorIdxJavaOpts}</value>
|
|
748
|
</property>
|
|
749
|
</configuration>
|
|
750
|
</sub-workflow>
|
|
751
|
<ok to="decision-documentssimilarity"/>
|
|
752
|
<error to="fail" />
|
|
753
|
</action>
|
|
754
|
|
|
755
|
<action name="skip-citationmatching">
|
|
756
|
<java>
|
|
757
|
<prepare>
|
|
758
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
759
|
<delete path="${nameNode}${workingDir}/citationmatching_chain" />
|
|
760
|
<mkdir path="${nameNode}${workingDir}/citationmatching_chain" />
|
|
761
|
</prepare>
|
|
762
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
763
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
764
|
<arg>-C{citation,
|
|
765
|
eu.dnetlib.iis.citationmatching.schemas.Citation,
|
|
766
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
767
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
768
|
<arg>-Ocitation=${workingDir}/citationmatching_chain/output</arg>
|
|
769
|
</java>
|
|
770
|
<ok to="decision-documentssimilarity"/>
|
|
771
|
<error to="fail"/>
|
|
772
|
</action>
|
|
773
|
<!-- end of citation matching part -->
|
|
774
|
|
|
775
|
<!-- start of documents similarity part -->
|
|
776
|
<!-- running documentsimilarity sequentially to all the other KDM modules
|
|
777
|
due to the lack of memory when executed in parallel -->
|
|
778
|
<decision name="decision-documentssimilarity">
|
|
779
|
<switch>
|
|
780
|
<case to="transformers_documentssimilarity">${active_documentssimilarity eq "true"}</case>
|
|
781
|
<default to="skip-documentssimilarity"/>
|
|
782
|
</switch>
|
|
783
|
</decision>
|
|
784
|
|
|
785
|
<action name="transformers_documentssimilarity">
|
|
786
|
<sub-workflow>
|
|
787
|
<app-path>${wf:appPath()}/transformers_documentssimilarity</app-path>
|
|
788
|
<propagate-configuration/>
|
|
789
|
<configuration>
|
|
790
|
<property>
|
|
791
|
<name>workingDir</name>
|
|
792
|
<value>${workingDir}/transformers_documentssimilarity/working_dir</value>
|
|
793
|
</property>
|
|
794
|
<property>
|
|
795
|
<name>input_person</name>
|
|
796
|
<value>${input_person}</value>
|
|
797
|
</property>
|
|
798
|
<property>
|
|
799
|
<name>input_metadata</name>
|
|
800
|
<value>${workingDir}/transformers_metadatamerger/output_merged_metadata</value>
|
|
801
|
</property>
|
|
802
|
<property>
|
|
803
|
<name>output_document_metadata</name>
|
|
804
|
<value>${workingDir}/transformers_documentssimilarity/output_document_metadata</value>
|
|
805
|
</property>
|
|
806
|
</configuration>
|
|
807
|
</sub-workflow>
|
|
808
|
<ok to="documentssimilarity_chain"/>
|
|
809
|
<error to="fail" />
|
|
810
|
</action>
|
|
811
|
|
|
812
|
<action name="documentssimilarity_chain">
|
|
813
|
<sub-workflow>
|
|
814
|
<app-path>${wf:appPath()}/documentssimilarity_chain</app-path>
|
|
815
|
<propagate-configuration/>
|
|
816
|
<configuration>
|
|
817
|
<property>
|
|
818
|
<name>workingDir</name>
|
|
819
|
<value>${workingDir}/documentssimilarity_chain/working_dir</value>
|
|
820
|
</property>
|
|
821
|
<property>
|
|
822
|
<name>input_document</name>
|
|
823
|
<value>${workingDir}/transformers_documentssimilarity/output_document_metadata</value>
|
|
824
|
</property>
|
|
825
|
<property>
|
|
826
|
<name>output_documents_similarity</name>
|
|
827
|
<value>${output_document_similarity}</value>
|
|
828
|
</property>
|
|
829
|
<property>
|
|
830
|
<name>parallel</name>
|
|
831
|
<value>${ds_parallel}</value>
|
|
832
|
</property>
|
|
833
|
<property>
|
|
834
|
<name>mapredChildJavaOpts</name>
|
|
835
|
<value>${ds_mapredChildJavaOpts}</value>
|
|
836
|
</property>
|
|
837
|
<property>
|
|
838
|
<name>sample</name>
|
|
839
|
<value>${ds_sample}</value>
|
|
840
|
</property>
|
|
841
|
<property>
|
|
842
|
<name>removal_rate</name>
|
|
843
|
<value>${ds_removal_rate}</value>
|
|
844
|
</property>
|
|
845
|
<property>
|
|
846
|
<name>removal_least_used</name>
|
|
847
|
<value>${ds_removal_least_used}</value>
|
|
848
|
</property>
|
|
849
|
<property>
|
|
850
|
<name>tfidfTopnTermPerDocument</name>
|
|
851
|
<value>${ds_tfidfTopnTermPerDocument}</value>
|
|
852
|
</property>
|
|
853
|
<property>
|
|
854
|
<name>similarityTopnDocumentPerDocument</name>
|
|
855
|
<value>${ds_similarityTopnDocumentPerDocument}</value>
|
|
856
|
</property>
|
|
857
|
</configuration>
|
|
858
|
</sub-workflow>
|
|
859
|
<ok to="decision-statistics"/>
|
|
860
|
<error to="fail" />
|
|
861
|
</action>
|
|
862
|
|
|
863
|
<action name="skip-documentssimilarity">
|
|
864
|
<java>
|
|
865
|
<prepare>
|
|
866
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
867
|
<delete path="${nameNode}${workingDir}/documentssimilarity_chain" />
|
|
868
|
<delete path="${nameNode}${output_document_similarity}" />
|
|
869
|
<mkdir path="${nameNode}${workingDir}/documentssimilarity_chain" />
|
|
870
|
<mkdir path="${nameNode}${output_document_similarity}" />
|
|
871
|
</prepare>
|
|
872
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
873
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
874
|
<arg>-C{documentssimilarity,
|
|
875
|
eu.dnetlib.iis.documentssimilarity.schemas.DocumentSimilarity,
|
|
876
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
877
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
878
|
<arg>-Odocumentssimilarity=${output_document_similarity}</arg>
|
|
879
|
</java>
|
|
880
|
<ok to="decision-statistics"/>
|
|
881
|
<error to="fail"/>
|
|
882
|
</action>
|
|
883
|
<!-- end of documents similarity part -->
|
|
884
|
|
|
885
|
<!-- statistics are calculated at the end, because they are taking two forked paths
|
|
886
|
outcome into account: transformers_metadatamerger and referenceextraction_project -->
|
|
887
|
<!-- statistics part -->
|
|
888
|
<decision name="decision-statistics">
|
|
889
|
<switch>
|
|
890
|
<case to="transformers_statistics">${active_statistics eq "true"}</case>
|
|
891
|
<default to="skip-statistics"/>
|
|
892
|
</switch>
|
|
893
|
</decision>
|
|
894
|
|
|
895
|
<action name="transformers_statistics">
|
|
896
|
<sub-workflow>
|
|
897
|
<app-path>${wf:appPath()}/transformers_statistics</app-path>
|
|
898
|
<propagate-configuration/>
|
|
899
|
<configuration>
|
|
900
|
<property>
|
|
901
|
<name>workingDir</name>
|
|
902
|
<value>${workingDir}/transformers_statistics/working_dir</value>
|
|
903
|
</property>
|
|
904
|
<property>
|
|
905
|
<name>input_document</name>
|
|
906
|
<value>${workingDir}/transformers_metadatamerger/output_merged_metadata</value>
|
|
907
|
</property>
|
|
908
|
<property>
|
|
909
|
<name>input_citation</name>
|
|
910
|
<value>${workingDir}/citationmatching_chain/output</value>
|
|
911
|
</property>
|
|
912
|
<property>
|
|
913
|
<!-- NOTICE: reference extraction will have to be enabled to get this input -->
|
|
914
|
<name>input_document_to_project</name>
|
|
915
|
<value>${output_document_to_project}</value>
|
|
916
|
</property>
|
|
917
|
<property>
|
|
918
|
<name>input_person</name>
|
|
919
|
<value>${input_person}</value>
|
|
920
|
</property>
|
|
921
|
<property>
|
|
922
|
<name>input_project</name>
|
|
923
|
<value>${input_project}</value>
|
|
924
|
</property>
|
|
925
|
<property>
|
|
926
|
<name>output_document_authors_citations</name>
|
|
927
|
<value>${workingDir}/transformers_statistics/output_document_authors_citations</value>
|
|
928
|
</property>
|
|
929
|
<property>
|
|
930
|
<name>output_person_id</name>
|
|
931
|
<value>${workingDir}/transformers_statistics/output_person_id</value>
|
|
932
|
</property>
|
|
933
|
<property>
|
|
934
|
<name>output_project_id</name>
|
|
935
|
<value>${workingDir}/transformers_statistics/output_project_id</value>
|
|
936
|
</property>
|
|
937
|
</configuration>
|
|
938
|
</sub-workflow>
|
|
939
|
<ok to="statistics"/>
|
|
940
|
<error to="fail" />
|
|
941
|
</action>
|
|
942
|
|
|
943
|
<action name="statistics">
|
|
944
|
<sub-workflow>
|
|
945
|
<app-path>${wf:appPath()}/statistics</app-path>
|
|
946
|
<propagate-configuration/>
|
|
947
|
<configuration>
|
|
948
|
<property>
|
|
949
|
<name>workingDir</name>
|
|
950
|
<value>${workingDir}/statistics/working_dir</value>
|
|
951
|
</property>
|
|
952
|
<property>
|
|
953
|
<name>input_document_authors_citations</name>
|
|
954
|
<value>${workingDir}/transformers_statistics/output_document_authors_citations</value>
|
|
955
|
</property>
|
|
956
|
<property>
|
|
957
|
<name>input_person_id</name>
|
|
958
|
<value>${workingDir}/transformers_statistics/output_person_id</value>
|
|
959
|
</property>
|
|
960
|
<property>
|
|
961
|
<name>input_project_id</name>
|
|
962
|
<value>${workingDir}/transformers_statistics/output_project_id</value>
|
|
963
|
</property>
|
|
964
|
<property>
|
|
965
|
<name>output_document_statistics</name>
|
|
966
|
<value>${output_document_statistics}</value>
|
|
967
|
</property>
|
|
968
|
<property>
|
|
969
|
<name>output_author_statistics</name>
|
|
970
|
<value>${output_author_statistics}</value>
|
|
971
|
</property>
|
|
972
|
<property>
|
|
973
|
<name>output_project_statistics</name>
|
|
974
|
<value>${output_project_statistics}</value>
|
|
975
|
</property>
|
|
976
|
<property>
|
|
977
|
<name>output_global_statistics</name>
|
|
978
|
<value>${workingDir}/statistics/output_global_statistics</value>
|
|
979
|
</property>
|
|
980
|
</configuration>
|
|
981
|
</sub-workflow>
|
|
982
|
<ok to="transformers_citations_from_matching"/>
|
|
983
|
<error to="fail" />
|
|
984
|
</action>
|
|
985
|
|
|
986
|
<action name="skip-statistics">
|
|
987
|
<java>
|
|
988
|
<prepare>
|
|
989
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
990
|
<delete path="${nameNode}${workingDir}/statistics" />
|
|
991
|
<delete path="${nameNode}${output_document_statistics}"/>
|
|
992
|
<delete path="${nameNode}${output_author_statistics}"/>
|
|
993
|
<delete path="${nameNode}${output_project_statistics}"/>
|
|
994
|
<mkdir path="${nameNode}${workingDir}/statistics" />
|
|
995
|
<mkdir path="${nameNode}${output_document_statistics}"/>
|
|
996
|
<mkdir path="${nameNode}${output_author_statistics}"/>
|
|
997
|
<mkdir path="${nameNode}${output_project_statistics}"/>
|
|
998
|
</prepare>
|
|
999
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
1000
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
1001
|
<arg>-C{document_statistics,
|
|
1002
|
eu.dnetlib.iis.statistics.schemas.DocumentToDocumentStatistics,
|
|
1003
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
1004
|
<arg>-C{author_statistics,
|
|
1005
|
eu.dnetlib.iis.statistics.schemas.AuthorToAuthorStatistics,
|
|
1006
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
1007
|
<arg>-C{project_statistics,
|
|
1008
|
eu.dnetlib.iis.statistics.schemas.ProjectToProjectStatistics,
|
|
1009
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
1010
|
<!-- FIXME currently global statistics are not available -->
|
|
1011
|
<!--
|
|
1012
|
<arg>-C{global_statistics,
|
|
1013
|
eu.dnetlib.iis.statistics.schemas.,
|
|
1014
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
1015
|
-->
|
|
1016
|
<!-- notice: directory have to aligned with skipped action output -->
|
|
1017
|
<arg>-Odocument_statistics=${output_document_statistics}</arg>
|
|
1018
|
<arg>-Oauthor_statistics=${output_author_statistics}</arg>
|
|
1019
|
<arg>-Oproject_statistics=${output_project_statistics}</arg>
|
|
1020
|
<!-- FIXME currently global statistics are not available -->
|
|
1021
|
<!--
|
|
1022
|
<arg>-Oglobal_statistics=${workingDir}/statistics/output_global_statistics</arg>
|
|
1023
|
-->
|
|
1024
|
</java>
|
|
1025
|
<ok to="transformers_citations_from_matching"/>
|
|
1026
|
<error to="fail"/>
|
|
1027
|
</action>
|
|
1028
|
<!-- end of statistics part -->
|
|
1029
|
|
|
1030
|
<!-- normalize and group citations part -->
|
|
1031
|
<action name="transformers_citations_from_matching">
|
|
1032
|
<sub-workflow>
|
|
1033
|
<app-path>${wf:appPath()}/transformers_citations_from_matching</app-path>
|
|
1034
|
<propagate-configuration/>
|
|
1035
|
<configuration>
|
|
1036
|
<property>
|
|
1037
|
<name>workingDir</name>
|
|
1038
|
<value>${workingDir}/transformers_citations_from_matching/working_dir</value>
|
|
1039
|
</property>
|
|
1040
|
<property>
|
|
1041
|
<name>input</name>
|
|
1042
|
<value>${workingDir}/citationmatching_chain/output</value>
|
|
1043
|
</property>
|
|
1044
|
<property>
|
|
1045
|
<name>output</name>
|
|
1046
|
<value>${workingDir}/transformers_citations_from_matching/output</value>
|
|
1047
|
</property>
|
|
1048
|
</configuration>
|
|
1049
|
</sub-workflow>
|
|
1050
|
<ok to="transformers_citations_from_ingestpmc"/>
|
|
1051
|
<error to="fail" />
|
|
1052
|
</action>
|
|
1053
|
|
|
1054
|
<action name="transformers_citations_from_ingestpmc">
|
|
1055
|
<sub-workflow>
|
|
1056
|
<app-path>${wf:appPath()}/transformers_citations_from_ingestpmc</app-path>
|
|
1057
|
<propagate-configuration/>
|
|
1058
|
<configuration>
|
|
1059
|
<property>
|
|
1060
|
<name>workingDir</name>
|
|
1061
|
<value>${workingDir}/transformers_citations_from_ingestpmc/working_dir</value>
|
|
1062
|
</property>
|
|
1063
|
<property>
|
|
1064
|
<name>input</name>
|
|
1065
|
<value>${input_citation_pmc}</value>
|
|
1066
|
</property>
|
|
1067
|
<property>
|
|
1068
|
<name>output</name>
|
|
1069
|
<value>${workingDir}/transformers_citations_from_ingestpmc/output</value>
|
|
1070
|
</property>
|
|
1071
|
</configuration>
|
|
1072
|
</sub-workflow>
|
|
1073
|
<ok to="citations_collapser"/>
|
|
1074
|
<error to="fail" />
|
|
1075
|
</action>
|
|
1076
|
|
|
1077
|
<action name="citations_collapser">
|
|
1078
|
<sub-workflow>
|
|
1079
|
<app-path>${wf:appPath()}/collapsers_multiple_input_collapser</app-path>
|
|
1080
|
<propagate-configuration/>
|
|
1081
|
<configuration>
|
|
1082
|
<property>
|
|
1083
|
<name>workingDir</name>
|
|
1084
|
<value>${workingDir}/citations_collapser/working_dir</value>
|
|
1085
|
</property>
|
|
1086
|
<!-- Input ports & parameters. -->
|
|
1087
|
<property>
|
|
1088
|
<name>origin_1</name>
|
|
1089
|
<value>ingested</value>
|
|
1090
|
</property>
|
|
1091
|
<property>
|
|
1092
|
<name>input_1</name>
|
|
1093
|
<value>${workingDir}/transformers_citations_from_ingestpmc/output</value>
|
|
1094
|
</property>
|
|
1095
|
<property>
|
|
1096
|
<name>origin_2</name>
|
|
1097
|
<value>matched</value>
|
|
1098
|
</property>
|
|
1099
|
<property>
|
|
1100
|
<name>input_2</name>
|
|
1101
|
<value>${workingDir}/transformers_citations_from_matching/output</value>
|
|
1102
|
</property>
|
|
1103
|
<property>
|
|
1104
|
<name>blocking_field</name>
|
|
1105
|
<value>sourceDocumentId</value>
|
|
1106
|
</property>
|
|
1107
|
<property>
|
|
1108
|
<name>schema_input</name>
|
|
1109
|
<value>eu.dnetlib.iis.common.citations.schemas.Citation</value>
|
|
1110
|
</property>
|
|
1111
|
<property>
|
|
1112
|
<name>output</name>
|
|
1113
|
<value>${workingDir}/citations_collapser/output</value>
|
|
1114
|
</property>
|
|
1115
|
<property>
|
|
1116
|
<name>schema_input_envelope</name>
|
|
1117
|
<value>eu.dnetlib.iis.common.citations.schemas.CitationEnvelope</value>
|
|
1118
|
</property>
|
|
1119
|
<property>
|
|
1120
|
<name>record_collapser</name>
|
|
1121
|
<value>eu.dnetlib.iis.collapsers.origins.PMCCitationCollapser</value>
|
|
1122
|
</property>
|
|
1123
|
</configuration>
|
|
1124
|
</sub-workflow>
|
|
1125
|
<ok to="transformers_export_citations"/>
|
|
1126
|
<error to="fail" />
|
|
1127
|
</action>
|
|
1128
|
|
|
1129
|
<action name="transformers_export_citations">
|
|
1130
|
<sub-workflow>
|
|
1131
|
<app-path>${wf:appPath()}/transformers_export_citations</app-path>
|
|
1132
|
<propagate-configuration/>
|
|
1133
|
<configuration>
|
|
1134
|
<property>
|
|
1135
|
<name>workingDir</name>
|
|
1136
|
<value>${workingDir}/transformers_export_citations/working_dir</value>
|
|
1137
|
</property>
|
|
1138
|
<property>
|
|
1139
|
<name>input</name>
|
|
1140
|
<value>${workingDir}/citations_collapser/output</value>
|
|
1141
|
</property>
|
|
1142
|
<property>
|
|
1143
|
<name>output</name>
|
|
1144
|
<value>${output_citation}</value>
|
|
1145
|
</property>
|
|
1146
|
</configuration>
|
|
1147
|
</sub-workflow>
|
|
1148
|
<ok to="end"/>
|
|
1149
|
<error to="fail" />
|
|
1150
|
</action>
|
|
1151
|
|
|
1152
|
<!-- end of normalize and group citations part -->
|
|
1153
|
|
|
1154
|
<kill name="fail">
|
|
1155
|
<message>Unfortunately, the process failed -- error message:
|
|
1156
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
1157
|
</kill>
|
|
1158
|
<end name="end" />
|
|
1159
|
</workflow-app>
|