1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="01ed11e8-e874-4478-a8ac-83e63e9699e4_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2016-06-25T10:24:32+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<CONFIGURATION>
|
11
|
|
12
|
<PARAMETERS>
|
13
|
|
14
|
<PARAM name="hbaseTable" description="HBase table name" required="true" type="string" />
|
15
|
<PARAM name="cluster" description="Hadoop cluster name" required="true" type="string" />
|
16
|
|
17
|
<PARAM name="dedupConfigSequence" description="dedup configuration orchestration name" required="true" type="string" />
|
18
|
<PARAM name="minDistWorkDir" description="work directory for the minDist algorithm" required="true" type="string" />
|
19
|
|
20
|
<PARAM name="mappingSimilarities" description="xslt mapping function for similarities" required="true" type="string" />
|
21
|
<PARAM name="mappingDissimilarities" description="xslt mapping function for dissimilarities" required="true" type="string" />
|
22
|
|
23
|
<PARAM name="mdFormat" description="dedup index mdFormat" required="true" type="string" default="OPENAIRE" />
|
24
|
<PARAM name="layout" description="dedup index layout" required="true" type="string" default="index" />
|
25
|
<PARAM name="interpretation" description="dedup index interpretation" required="true" type="string" default="dedup" />
|
26
|
|
27
|
</PARAMETERS>
|
28
|
|
29
|
<WORKFLOW>
|
30
|
|
31
|
<NODE isStart="true" name="fetchRelClasses" type="FetchRelClasses">
|
32
|
<DESCRIPTION/>
|
33
|
<PARAMETERS>
|
34
|
<PARAM name="xquery" property="dnet.openaire.model.relclasses.xquery" />
|
35
|
<PARAM name="relClassesName" value="relClasses" />
|
36
|
</PARAMETERS>
|
37
|
<ARCS>
|
38
|
<ARC to="resetPath"/>
|
39
|
</ARCS>
|
40
|
</NODE>
|
41
|
|
42
|
<NODE isStart="true" name="setDedupConfigs" type="SetDedupConfiguration">
|
43
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
44
|
<PARAMETERS>
|
45
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
|
46
|
<PARAM name="dedupConfigSequenceParam" value="dedup.conf.queue" />
|
47
|
<PARAM name="entityTypeNameParam" value="entityType" />
|
48
|
<PARAM name="entityTypeIdParam" value="entityTypeId" />
|
49
|
</PARAMETERS>
|
50
|
<ARCS>
|
51
|
<ARC to="resetPath"/>
|
52
|
</ARCS>
|
53
|
</NODE>
|
54
|
|
55
|
|
56
|
<NODE name="resetPath" type="CreateHdfsDirJob" isJoin="true">
|
57
|
<DESCRIPTION>input files cleanup</DESCRIPTION>
|
58
|
<PARAMETERS>
|
59
|
<PARAM name="force" value="true" />
|
60
|
<PARAM name="path" ref="minDistWorkDir" />
|
61
|
<PARAM name="cluster" ref="cluster" />
|
62
|
</PARAMETERS>
|
63
|
<ARCS>
|
64
|
<ARC to="deduplicateScan"/>
|
65
|
</ARCS>
|
66
|
</NODE>
|
67
|
|
68
|
|
69
|
<NODE name="deduplicateScan" type="DuplicateScanJob">
|
70
|
<DESCRIPTION>Dup Scan</DESCRIPTION>
|
71
|
<PARAMETERS>
|
72
|
<PARAM name="hadoopJob" value="dedupCandidateScanJob" />
|
73
|
<PARAM name="dedupConfigurationOrchestration" env="dedup.conf.queue" />
|
74
|
<PARAM name="jobParams">
|
75
|
<MAP>
|
76
|
<ENTRY key="entityTypeId" env="entityTypeId"/>
|
77
|
<ENTRY key="entityType" env="entityType"/>
|
78
|
<ENTRY key="cluster" ref="cluster"/>
|
79
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
80
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
81
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
82
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
83
|
</MAP>
|
84
|
</PARAM>
|
85
|
</PARAMETERS>
|
86
|
<ARCS>
|
87
|
<ARC to="deduplicateScan"/>
|
88
|
<ARC name="done" to="prepareActionSet"/>
|
89
|
<ARC name="done" to="queryUserSimilarities"/>
|
90
|
</ARCS>
|
91
|
</NODE>
|
92
|
|
93
|
<NODE name="prepareActionSet" type="PrepareConfiguredActionSet">
|
94
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
95
|
<PARAMETERS>
|
96
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
|
97
|
<PARAM name="jobProperty" value="rawSetId" />
|
98
|
<PARAM name="actionSetPathParam" value="actionSetPath" />
|
99
|
<PARAM name="setsParam" value="sets" />
|
100
|
</PARAMETERS>
|
101
|
<ARCS>
|
102
|
<ARC to="similarity2actions"/>
|
103
|
</ARCS>
|
104
|
</NODE>
|
105
|
|
106
|
<NODE name="similarity2actions" type="DedupSimilarityToActions">
|
107
|
<DESCRIPTION>export the similarity rels as Actions</DESCRIPTION>
|
108
|
<PARAMETERS>
|
109
|
<PARAM name="hadoopJob" value="dedupSimilarity2HdfsActionsJob" />
|
110
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
|
111
|
<PARAM name="entityType" env="entityType" />
|
112
|
<PARAM name="jobParams">
|
113
|
<MAP>
|
114
|
<ENTRY key="dedup.conf" env="dedup.conf"/>
|
115
|
<ENTRY key="entityTypeId" env="entityTypeId"/>
|
116
|
<ENTRY key="entityType" env="entityType"/>
|
117
|
<ENTRY key="cluster" ref="cluster"/>
|
118
|
<ENTRY key="rawSetId" env="rawSetId"/>
|
119
|
<ENTRY key="mapred.output.dir" env="actionSetPath"/>
|
120
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
121
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
122
|
</MAP>
|
123
|
</PARAM>
|
124
|
</PARAMETERS>
|
125
|
<ARCS>
|
126
|
<ARC to="updateActionSets"/>
|
127
|
</ARCS>
|
128
|
</NODE>
|
129
|
|
130
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
131
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
132
|
<PARAMETERS/>
|
133
|
<ARCS>
|
134
|
<ARC to="doneActions"/>
|
135
|
</ARCS>
|
136
|
</NODE>
|
137
|
|
138
|
<NODE name="queryUserSimilarities" type="QueryUserActionDbJob">
|
139
|
<DESCRIPTION>query user similarity</DESCRIPTION>
|
140
|
<PARAMETERS>
|
141
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
|
142
|
<PARAM name="db" property="dnet.dedup.db.name" />
|
143
|
<PARAM name="sql" value="/eu/dnetlib/msro/workflows/dedup/querySimilaritiesBySet.sql.st" />
|
144
|
<PARAM name="outputEprParam" value="simEpr" />
|
145
|
</PARAMETERS>
|
146
|
<ARCS>
|
147
|
<ARC to="buildSimilarityMesh"/>
|
148
|
</ARCS>
|
149
|
</NODE>
|
150
|
|
151
|
<NODE name="buildSimilarityMesh" type="BuildSimilarityMeshJob">
|
152
|
<DESCRIPTION>build mesh</DESCRIPTION>
|
153
|
<PARAMETERS>
|
154
|
<PARAM name="inputEprParam" value="simEpr" />
|
155
|
<PARAM name="outputEprParam" value="simMeshEpr" />
|
156
|
</PARAMETERS>
|
157
|
<ARCS>
|
158
|
<ARC to="storeSimilarities"/>
|
159
|
</ARCS>
|
160
|
</NODE>
|
161
|
|
162
|
<NODE name="storeSimilarities" type="StoreHBase">
|
163
|
<DESCRIPTION>store similarity</DESCRIPTION>
|
164
|
<PARAMETERS>
|
165
|
<PARAM name="inputEprParam" value="simMeshEpr" />
|
166
|
<PARAM name="hbaseTable" ref="hbaseTable" />
|
167
|
<PARAM name="cluster" ref="cluster" />
|
168
|
<PARAM name="mapping" ref="mappingSimilarities" />
|
169
|
<PARAM name="simulation" value="false" />
|
170
|
</PARAMETERS>
|
171
|
<ARCS>
|
172
|
<ARC to="queryUserDissimilarities"/>
|
173
|
</ARCS>
|
174
|
</NODE>
|
175
|
|
176
|
<NODE name="queryUserDissimilarities" type="QueryUserActionDbJob">
|
177
|
<DESCRIPTION>query user dissimilarity</DESCRIPTION>
|
178
|
<PARAMETERS>
|
179
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
|
180
|
<PARAM name="db" property="dnet.dedup.db.name" />
|
181
|
<PARAM name="sql" value="/eu/dnetlib/msro/workflows/dedup/queryDissimilaritiesBySet.sql.st" />
|
182
|
<PARAM name="outputEprParam" value="dissimEpr" />
|
183
|
</PARAMETERS>
|
184
|
<ARCS>
|
185
|
<ARC to="storeDissimilarities"/>
|
186
|
</ARCS>
|
187
|
</NODE>
|
188
|
|
189
|
<NODE name="storeDissimilarities" type="DeleteFromHBase">
|
190
|
<DESCRIPTION>store dissimilarity</DESCRIPTION>
|
191
|
<PARAMETERS>
|
192
|
<PARAM name="inputEprParam" value="dissimEpr" />
|
193
|
<PARAM name="hbaseTable" ref="hbaseTable" />
|
194
|
<PARAM name="cluster" ref="cluster" />
|
195
|
<PARAM name="mapping" ref="mappingDissimilarities" />
|
196
|
<PARAM name="simulation" value="false" />
|
197
|
</PARAMETERS>
|
198
|
<ARCS>
|
199
|
<ARC to="doneActions"/>
|
200
|
</ARCS>
|
201
|
</NODE>
|
202
|
|
203
|
<NODE name="doneActions" isJoin="true">
|
204
|
<DESCRIPTION>done actions</DESCRIPTION>
|
205
|
<PARAMETERS/>
|
206
|
<ARCS>
|
207
|
<ARC to="closeMesh"/>
|
208
|
</ARCS>
|
209
|
</NODE>
|
210
|
|
211
|
<NODE name="closeMesh" type="LaunchWorkflowTemplate">
|
212
|
<DESCRIPTION>close mesh</DESCRIPTION>
|
213
|
<PARAMETERS>
|
214
|
<PARAM name="wfTemplateId" value="9522876c-5885-4dd2-9d06-92e9a3eaa9f1_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ==" />
|
215
|
<PARAM name="wfTemplateParams">
|
216
|
<MAP>
|
217
|
<ENTRY key="hbaseTable" ref="hbaseTable"/>
|
218
|
<ENTRY key="cluster" ref="cluster"/>
|
219
|
<ENTRY key="workDir" ref="minDistWorkDir"/>
|
220
|
<ENTRY key="entitySequence" env="entityType"/>
|
221
|
<ENTRY key="dedupConfigSequence" ref="dedupConfigSequence" />
|
222
|
</MAP>
|
223
|
</PARAM>
|
224
|
</PARAMETERS>
|
225
|
<ARCS>
|
226
|
<ARC to="success" />
|
227
|
</ARCS>
|
228
|
</NODE>
|
229
|
|
230
|
</WORKFLOW>
|
231
|
</CONFIGURATION>
|
232
|
</BODY>
|
233
|
</RESOURCE_PROFILE>
|