1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="70274106-375d-4135-9de1-536a606b327b_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2017-02-28T10:21:23+01:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<CONFIGURATION>
|
11
|
<PARAMETERS>
|
12
|
<PARAM description="HBase table name" name="hbaseTable" required="true" type="string"/>
|
13
|
<PARAM description="Hadoop cluster name" name="cluster" required="true" type="string"/>
|
14
|
<PARAM description="minDist algorithm work directory" name="workDir" required="true" type="string"/>
|
15
|
<PARAM description="csv list of action sets to be promoted" name="actionSets" required="true" type="string"/>
|
16
|
<PARAM default="organization,result" description="list of entity names to be considered by the algorithm" name="entitySequence" required="true" type="string"/>
|
17
|
</PARAMETERS>
|
18
|
<WORKFLOW>
|
19
|
<NODE isStart="true" name="fetchOntologies" type="FetchOntologies">
|
20
|
<DESCRIPTION/>
|
21
|
<PARAMETERS>
|
22
|
<PARAM name="ontologiesParamName" value="ontologies"/>
|
23
|
</PARAMETERS>
|
24
|
<ARCS>
|
25
|
<ARC to="promoteActions"/>
|
26
|
</ARCS>
|
27
|
</NODE>
|
28
|
<NODE name="promoteActions" type="PromoteActionsHDFS">
|
29
|
<DESCRIPTION>Promote Actions</DESCRIPTION>
|
30
|
<PARAMETERS>
|
31
|
<PARAM name="set" ref="actionSets"/>
|
32
|
<PARAM name="tableName" ref="hbaseTable"/>
|
33
|
</PARAMETERS>
|
34
|
<ARCS>
|
35
|
<ARC to="entitySequence"/>
|
36
|
</ARCS>
|
37
|
</NODE>
|
38
|
<NODE name="entitySequence" type="CheckEntitySequenceJob">
|
39
|
<DESCRIPTION>Check entity sequence</DESCRIPTION>
|
40
|
<PARAMETERS>
|
41
|
<PARAM name="entitySequence" ref="entitySequence"/>
|
42
|
<PARAM name="dedupConfigSequenceParam" value="dedup.conf.queue"/>
|
43
|
<PARAM name="entityTypeNameParam" value="dedup.entity.name"/>
|
44
|
<PARAM name="entityTypeIdParam" value="dedup.entity.id"/>
|
45
|
<PARAM name="actionSetParam" value="actionset"/>
|
46
|
</PARAMETERS>
|
47
|
<ARCS>
|
48
|
<ARC to="resetPath"/>
|
49
|
<ARC name="done" to="success"/>
|
50
|
</ARCS>
|
51
|
</NODE>
|
52
|
<NODE name="resetPath" type="CreateHdfsDirJob">
|
53
|
<DESCRIPTION>input files cleanup</DESCRIPTION>
|
54
|
<PARAMETERS>
|
55
|
<PARAM name="force" value="true"/>
|
56
|
<PARAM name="path" ref="workDir"/>
|
57
|
<PARAM name="cluster" ref="cluster"/>
|
58
|
</PARAMETERS>
|
59
|
<ARCS>
|
60
|
<ARC to="mindist"/>
|
61
|
</ARCS>
|
62
|
</NODE>
|
63
|
<NODE name="mindist" type="MinDistSearchHadoopJob">
|
64
|
<DESCRIPTION>find the minimum vertex in each adjacency lists</DESCRIPTION>
|
65
|
<PARAMETERS>
|
66
|
<PARAM name="debug" value="false"/>
|
67
|
<PARAM name="cluster" ref="cluster"/>
|
68
|
<PARAM name="outPathParam" value="outputPath"/>
|
69
|
<PARAM env="actionset" name="dedupConfigSequence"/>
|
70
|
<PARAM name="workDir" ref="workDir"/>
|
71
|
<PARAM name="jobParams">
|
72
|
<MAP>
|
73
|
<ENTRY key="cluster" ref="cluster"/>
|
74
|
<ENTRY env="dedup.entity.id" key="entityTypeId"/>
|
75
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
76
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
77
|
</MAP>
|
78
|
</PARAM>
|
79
|
</PARAMETERS>
|
80
|
<ARCS>
|
81
|
<ARC name="depth_n" to="mindist"/>
|
82
|
<ARC to="components"/>
|
83
|
</ARCS>
|
84
|
</NODE>
|
85
|
<NODE name="components" type="DedupConfigurationLoaderJob">
|
86
|
<DESCRIPTION>joins all the vertex ids to build the connected components in the graph</DESCRIPTION>
|
87
|
<PARAMETERS>
|
88
|
<PARAM name="hadoopJob" value="connectedComponentsJob"/>
|
89
|
<PARAM env="actionset" name="dedupConfigSequence"/>
|
90
|
<PARAM name="cluster" ref="cluster"/>
|
91
|
<PARAM name="jobParams">
|
92
|
<MAP>
|
93
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
94
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
95
|
<ENTRY env="dedup.entity.name" key="entityType"/>
|
96
|
<ENTRY env="dedup.entity.id" key="entityTypeId"/>
|
97
|
<ENTRY env="outputPath" key="mapred.input.dir"/>
|
98
|
</MAP>
|
99
|
</PARAM>
|
100
|
</PARAMETERS>
|
101
|
<ARCS>
|
102
|
<ARC to="markDeleted"/>
|
103
|
</ARCS>
|
104
|
</NODE>
|
105
|
<NODE name="markDeleted" type="DedupConfigurationLoaderJob">
|
106
|
<DESCRIPTION>mark duplicates as deleted by inference</DESCRIPTION>
|
107
|
<PARAMETERS>
|
108
|
<PARAM name="hadoopJob" value="dedupMarkDeletedEntityJob"/>
|
109
|
<PARAM env="actionset" name="dedupConfigSequence"/>
|
110
|
<PARAM name="cluster" ref="cluster"/>
|
111
|
<PARAM name="jobParams">
|
112
|
<MAP>
|
113
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
114
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
115
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
116
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
117
|
<ENTRY env="dedup.entity.name" key="entityType"/>
|
118
|
<ENTRY env="dedup.entity.id" key="entityTypeId"/>
|
119
|
</MAP>
|
120
|
</PARAM>
|
121
|
</PARAMETERS>
|
122
|
<ARCS>
|
123
|
<ARC to="buildRoots"/>
|
124
|
</ARCS>
|
125
|
</NODE>
|
126
|
<NODE name="buildRoots" type="DedupConfigurationLoaderJob">
|
127
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
128
|
<PARAMETERS>
|
129
|
<PARAM name="hadoopJob" value="dedupBuildRootsJob"/>
|
130
|
<PARAM env="actionset" name="dedupConfigSequence"/>
|
131
|
<PARAM name="cluster" ref="cluster"/>
|
132
|
<PARAM name="jobParams">
|
133
|
<MAP>
|
134
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
135
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
136
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
137
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
138
|
<ENTRY env="dedup.entity.name" key="entityType"/>
|
139
|
<ENTRY env="dedup.entity.id" key="entityTypeId"/>
|
140
|
<ENTRY env="ontologies" key="ontologies"/>
|
141
|
</MAP>
|
142
|
</PARAM>
|
143
|
</PARAMETERS>
|
144
|
<ARCS>
|
145
|
<ARC to="mergeRels"/>
|
146
|
</ARCS>
|
147
|
</NODE>
|
148
|
<NODE name="mergeRels" type="DedupConfigurationLoaderJob">
|
149
|
<DESCRIPTION>merge rels</DESCRIPTION>
|
150
|
<PARAMETERS>
|
151
|
<PARAM name="hadoopJob" value="mergeRelationVersionJob"/>
|
152
|
<PARAM env="actionset" name="dedupConfigSequence"/>
|
153
|
<PARAM name="cluster" ref="cluster"/>
|
154
|
<PARAM name="jobParams">
|
155
|
<MAP>
|
156
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
157
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
158
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
159
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
160
|
<ENTRY env="dedup.entity.name" key="entityType"/>
|
161
|
<ENTRY env="dedup.entity.id" key="entityTypeId"/>
|
162
|
<ENTRY env="ontologies" key="ontologies"/>
|
163
|
</MAP>
|
164
|
</PARAM>
|
165
|
</PARAMETERS>
|
166
|
<ARCS>
|
167
|
<ARC to="entitySequence"/>
|
168
|
</ARCS>
|
169
|
</NODE>
|
170
|
</WORKFLOW>
|
171
|
</CONFIGURATION>
|
172
|
</BODY>
|
173
|
</RESOURCE_PROFILE>
|