1 |
35866
|
claudio.at
|
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
|
|
<RESOURCE_PROFILE xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
3 |
|
|
<HEADER>
|
4 |
|
|
<RESOURCE_IDENTIFIER value="ab5d7de1-b23e-495a-9928-be62a6bbd8b6_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
5 |
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
6 |
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
7 |
|
|
<RESOURCE_URI value=""/>
|
8 |
|
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
9 |
|
|
</HEADER>
|
10 |
|
|
<BODY>
|
11 |
|
|
<WORKFLOW_NAME>Offline Deduplication</WORKFLOW_NAME>
|
12 |
|
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
13 |
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14 |
|
|
<CONFIGURATION start="manual">
|
15 |
|
|
<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true">
|
16 |
|
|
<DESCRIPTION />
|
17 |
|
|
<PARAMETERS>
|
18 |
|
|
<PARAM name="relClassesProperty" type="string" managedBy="system" required="true">dnet.openaire.model.relclasses.xquery</PARAM>
|
19 |
|
|
<PARAM name="relClassesName" type="string" managedBy="system" required="true">relClasses</PARAM>
|
20 |
|
|
</PARAMETERS>
|
21 |
|
|
<ARCS>
|
22 |
|
|
<ARC to="SELECT_MODE" />
|
23 |
|
|
</ARCS>
|
24 |
|
|
</NODE>
|
25 |
|
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
26 |
|
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
27 |
|
|
<PARAMETERS>
|
28 |
36623
|
claudio.at
|
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM>
|
29 |
35866
|
claudio.at
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
30 |
|
|
</PARAMETERS>
|
31 |
|
|
<ARCS>
|
32 |
|
|
<ARC to="SELECT_MODE" />
|
33 |
|
|
</ARCS>
|
34 |
|
|
</NODE>
|
35 |
|
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
36 |
|
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
37 |
|
|
<PARAMETERS>
|
38 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
39 |
|
|
<PARAM required="true" type="string" name="table" managedBy="user">db_stdl</PARAM>
|
40 |
|
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
41 |
|
|
</PARAMETERS>
|
42 |
|
|
<ARCS>
|
43 |
|
|
<ARC to="SELECT_MODE" />
|
44 |
|
|
</ARCS>
|
45 |
|
|
</NODE>
|
46 |
|
|
<NODE name="SELECT_MODE" type="Selection" isJoin="true">
|
47 |
|
|
<DESCRIPTION>Select Workflow execution MODE.</DESCRIPTION>
|
48 |
|
|
<PARAMETERS>
|
49 |
|
|
<PARAM function="validValues(['DUPLICATE_SCAN', 'CLOSE_MESH', 'EXPORT_CSV', 'EXPORT_ACTIONS'])" managedBy="user" name="selection" required="true" type="string"></PARAM>
|
50 |
|
|
</PARAMETERS>
|
51 |
|
|
<ARCS>
|
52 |
|
|
<ARC name="DUPLICATE_SCAN" to="deduplicateScan" />
|
53 |
|
|
<ARC name="CLOSE_MESH" to="dedupGrouper" />
|
54 |
|
|
<ARC name="EXPORT_CSV" to="setCsvPath" />
|
55 |
36643
|
claudio.at
|
<ARC name="EXPORT_ACTIONS" to="prepareActionSet" />
|
56 |
35866
|
claudio.at
|
</ARCS>
|
57 |
|
|
</NODE>
|
58 |
|
|
|
59 |
|
|
<NODE name="deduplicateScan" type="DuplicateScanJob">
|
60 |
|
|
<DESCRIPTION>Dup Scan</DESCRIPTION>
|
61 |
|
|
<PARAMETERS>
|
62 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupCandidateScanJob</PARAM>
|
63 |
|
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
64 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
65 |
|
|
{
|
66 |
|
|
'entityTypeId' : 'entityTypeId',
|
67 |
|
|
'entityType' : 'entityType',
|
68 |
|
|
'cluster' : 'cluster',
|
69 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
70 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
71 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
72 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
73 |
|
|
}
|
74 |
|
|
</PARAM>
|
75 |
|
|
</PARAMETERS>
|
76 |
|
|
<ARCS>
|
77 |
|
|
<ARC to="deduplicateScan" />
|
78 |
|
|
<ARC name="done" to="success" />
|
79 |
|
|
</ARCS>
|
80 |
|
|
</NODE>
|
81 |
|
|
|
82 |
|
|
<NODE name="dedupGrouper" type="DedupGrouperJob">
|
83 |
|
|
<DESCRIPTION>dedup grouper</DESCRIPTION>
|
84 |
|
|
<PARAMETERS>
|
85 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupGrouperJob</PARAM>
|
86 |
|
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
87 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
88 |
|
|
{
|
89 |
36623
|
claudio.at
|
'dedup.conf' : 'dedup.conf',
|
90 |
35866
|
claudio.at
|
'entityTypeId' : 'entityTypeId',
|
91 |
|
|
'entityType' : 'entityType',
|
92 |
|
|
'cluster' : 'cluster',
|
93 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
94 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
95 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
96 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
97 |
|
|
}
|
98 |
|
|
</PARAM>
|
99 |
|
|
</PARAMETERS>
|
100 |
|
|
<ARCS>
|
101 |
|
|
<ARC to="dedupGrouper" />
|
102 |
|
|
<ARC name="done" to="findRoots" />
|
103 |
|
|
</ARCS>
|
104 |
|
|
</NODE>
|
105 |
|
|
<NODE name="findRoots" type="SubmitHadoopJob">
|
106 |
|
|
<DESCRIPTION>find roots</DESCRIPTION>
|
107 |
|
|
<PARAMETERS>
|
108 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupFindRootsJob</PARAM>
|
109 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
110 |
|
|
{
|
111 |
36623
|
claudio.at
|
'dedup.conf' : 'dedup.conf',
|
112 |
35866
|
claudio.at
|
'entityTypeId' : 'entityTypeId',
|
113 |
|
|
'entityType' : 'entityType',
|
114 |
|
|
'cluster' : 'cluster',
|
115 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
116 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
117 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
118 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
119 |
|
|
}
|
120 |
|
|
</PARAM>
|
121 |
|
|
</PARAMETERS>
|
122 |
|
|
<ARCS>
|
123 |
|
|
<ARC to="buildRoots" />
|
124 |
|
|
</ARCS>
|
125 |
|
|
</NODE>
|
126 |
|
|
<NODE name="buildRoots" type="SubmitHadoopJob">
|
127 |
|
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
128 |
|
|
<PARAMETERS>
|
129 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupBuildRootsJob</PARAM>
|
130 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
131 |
|
|
{
|
132 |
36623
|
claudio.at
|
'dedup.conf' : 'dedup.conf',
|
133 |
35866
|
claudio.at
|
'relClasses' : 'relClasses',
|
134 |
|
|
'entityTypeId' : 'entityTypeId',
|
135 |
|
|
'entityType' : 'entityType',
|
136 |
|
|
'cluster' : 'cluster',
|
137 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
138 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
139 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
140 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
141 |
|
|
}
|
142 |
|
|
</PARAM>
|
143 |
|
|
</PARAMETERS>
|
144 |
|
|
<ARCS>
|
145 |
|
|
<ARC to="success" />
|
146 |
|
|
</ARCS>
|
147 |
|
|
</NODE>
|
148 |
|
|
|
149 |
|
|
<NODE name="setCsvPath" type="SetEnvParameter">
|
150 |
|
|
<DESCRIPTION>Set the CSV file path on HDFS</DESCRIPTION>
|
151 |
|
|
<PARAMETERS>
|
152 |
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">csvPath</PARAM>
|
153 |
|
|
<PARAM managedBy="user" name="parameterValue" required="false" type="string"></PARAM>
|
154 |
|
|
</PARAMETERS>
|
155 |
|
|
<ARCS>
|
156 |
|
|
<ARC to="cleanupCsv" />
|
157 |
|
|
</ARCS>
|
158 |
|
|
</NODE>
|
159 |
|
|
|
160 |
|
|
<NODE name="cleanupCsv" type="DeleteHdfsPathJob">
|
161 |
|
|
<DESCRIPTION>CSV files cleanup</DESCRIPTION>
|
162 |
|
|
<PARAMETERS>
|
163 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
164 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
165 |
|
|
{
|
166 |
|
|
'path' : 'csvPath'
|
167 |
|
|
}
|
168 |
|
|
</PARAM>
|
169 |
|
|
</PARAMETERS>
|
170 |
|
|
<ARCS>
|
171 |
|
|
<ARC to="roots2CSV" />
|
172 |
|
|
</ARCS>
|
173 |
|
|
</NODE>
|
174 |
|
|
|
175 |
36765
|
claudio.at
|
<NODE name="roots2CSV" type="DedupConfigurationAwareJobLoader">
|
176 |
35866
|
claudio.at
|
<DESCRIPTION>export the representative entities as CSV files</DESCRIPTION>
|
177 |
|
|
<PARAMETERS>
|
178 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupRootsToCSVJob</PARAM>
|
179 |
36797
|
claudio.at
|
<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM>
|
180 |
35866
|
claudio.at
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
181 |
|
|
{
|
182 |
36623
|
claudio.at
|
'dedup.conf' : 'dedup.conf',
|
183 |
35866
|
claudio.at
|
'relClasses' : 'relClasses',
|
184 |
|
|
'entityTypeId' : 'entityTypeId',
|
185 |
|
|
'entityType' : 'entityType',
|
186 |
|
|
'cluster' : 'cluster',
|
187 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
188 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
189 |
|
|
'mapred.output.dir' : 'csvPath'
|
190 |
|
|
}
|
191 |
|
|
</PARAM>
|
192 |
|
|
</PARAMETERS>
|
193 |
|
|
<ARCS>
|
194 |
|
|
<ARC to="success" />
|
195 |
|
|
</ARCS>
|
196 |
|
|
</NODE>
|
197 |
|
|
|
198 |
36643
|
claudio.at
|
<NODE name="prepareActionSet" type="PrepareConfiguredActionSet">
|
199 |
35866
|
claudio.at
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
200 |
|
|
<PARAMETERS>
|
201 |
36643
|
claudio.at
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
202 |
|
|
<PARAM required="true" type="string" name="jobProperty" managedBy="system">rawSetId</PARAM>
|
203 |
35866
|
claudio.at
|
</PARAMETERS>
|
204 |
|
|
<ARCS>
|
205 |
|
|
<ARC to="similarity2actions" />
|
206 |
|
|
</ARCS>
|
207 |
|
|
</NODE>
|
208 |
|
|
|
209 |
|
|
<NODE name="similarity2actions" type="DedupSimilarityToActionsJobNode">
|
210 |
|
|
<DESCRIPTION>export the similarity rels as Actions</DESCRIPTION>
|
211 |
|
|
<PARAMETERS>
|
212 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupSimilarity2ActionsJob</PARAM>
|
213 |
|
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
214 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
215 |
|
|
{
|
216 |
36623
|
claudio.at
|
'dedup.conf' : 'dedup.conf',
|
217 |
35866
|
claudio.at
|
'entityTypeId' : 'entityTypeId',
|
218 |
|
|
'entityType' : 'entityType',
|
219 |
|
|
'cluster' : 'cluster',
|
220 |
|
|
'rawSetId' : 'rawSetId',
|
221 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
222 |
|
|
'hbase.mapreduce.inputtable' : 'tableName'
|
223 |
|
|
}
|
224 |
|
|
</PARAM>
|
225 |
|
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
226 |
|
|
{
|
227 |
|
|
'hbase.mapred.outputtable' : 'hbase.actions.table',
|
228 |
|
|
'hbase.mapreduce.outputtable' : 'hbase.actions.table'
|
229 |
|
|
}
|
230 |
|
|
</PARAM>
|
231 |
|
|
</PARAMETERS>
|
232 |
|
|
<ARCS>
|
233 |
|
|
<ARC to="updateActionSets" />
|
234 |
|
|
</ARCS>
|
235 |
|
|
</NODE>
|
236 |
|
|
|
237 |
|
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
238 |
|
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
239 |
|
|
<PARAMETERS/>
|
240 |
|
|
<ARCS>
|
241 |
|
|
<ARC to="success" />
|
242 |
|
|
</ARCS>
|
243 |
|
|
</NODE>
|
244 |
|
|
|
245 |
|
|
</CONFIGURATION>
|
246 |
|
|
<STATUS />
|
247 |
|
|
</BODY>
|
248 |
|
|
</RESOURCE_PROFILE>
|