Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER value="9522876c-5885-4dd2-9d06-92e9a3eaa9f1_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
4
		<RESOURCE_TYPE value="WorkflowTemplateDSResourceType" />
5
		<RESOURCE_KIND value="WorkflowTemplateDSResources" />
6
		<RESOURCE_URI value=""/>
7
		<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
8
	</HEADER>
9
	<BODY>
10
		<CONFIGURATION>
11

    
12
			<PARAMETERS>
13
				<PARAM name="hbaseTable"        description="HBase table name" required="true" type="string" />
14
				<PARAM name="cluster"           description="Hadoop cluster name" required="true" type="string" />
15
				<PARAM name="workDir"           description="minDist algorithm work directory" required="true" type="string" />
16
				<PARAM name="entitySequence"    description="list of entity names to be considered by the algorithm" required="true" type="string" default="organization,result" />
17
				<PARAM name="dedupConfigSequence" description="dedup configuration orchestration name" required="true" type="string" />
18
			</PARAMETERS>
19

    
20
			<WORKFLOW>
21

    
22
				<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true" >
23
					<DESCRIPTION/>
24
					<PARAMETERS>
25
						<PARAM name="xquery" property="dnet.openaire.model.relclasses.xquery" />
26
						<PARAM name="relClassesName" value="relClasses" />
27
					</PARAMETERS>
28
					<ARCS>
29
						<ARC to="entitySequence"/>
30
					</ARCS>
31
				</NODE>
32

    
33
				<NODE name="entitySequence" type="CheckEntitySequenceJob">
34
					<DESCRIPTION>Check entity sequence</DESCRIPTION>
35
					<PARAMETERS>
36
						<PARAM name="entitySequence" ref="entitySequence" />
37
						<PARAM name="dedupConfigSequenceParam" value="dedup.conf.queue" />
38
						<PARAM name="entityTypeNameParam" value="dedup.entity.name" />
39
						<PARAM name="entityTypeIdParam" value="dedup.entity.id" />
40
					</PARAMETERS>
41
					<ARCS>
42
						<ARC to="resetPath"/>
43
						<ARC name="done" to="success"/>
44
					</ARCS>
45
				</NODE>
46

    
47
				<NODE name="resetPath" type="CreateHdfsDirJob">
48
					<DESCRIPTION>input files cleanup</DESCRIPTION>
49
					<PARAMETERS>
50
						<PARAM name="force" value="true" />
51
						<PARAM name="path" ref="workDir" />
52
						<PARAM name="cluster" ref="cluster" />
53
					</PARAMETERS>
54
					<ARCS>
55
						<ARC to="mindist"/>
56
					</ARCS>
57
				</NODE>
58

    
59
				<NODE name="mindist" type="MinDistSearchHadoopJob">
60
					<DESCRIPTION>find the minimum vertex in each adjacency lists</DESCRIPTION>
61
					<PARAMETERS>
62
						<PARAM name="debug" value="false" />
63
						<PARAM name="cluster" ref="cluster" />
64
						<PARAM name="outPathParam" value="outputPath" />
65
						<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
66
						<PARAM name="workDir" ref="workDir"/>
67
						<PARAM name="jobParams">
68
							<MAP>
69
								<ENTRY key="cluster" ref="cluster"/>
70
								<ENTRY key="entityTypeId" env="dedup.entity.id"/>
71
								<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
72
								<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
73
							</MAP>
74
						</PARAM>
75
					</PARAMETERS>
76
					<ARCS>
77
						<ARC name="depth_n" to="mindist"/>
78
						<ARC to="components"/>
79
					</ARCS>
80
				</NODE>
81

    
82
				<NODE name="components" type="DedupConfigurationLoaderJob">
83
					<DESCRIPTION>joins all the vertex ids to build the connected components in the graph</DESCRIPTION>
84
					<PARAMETERS>
85
						<PARAM name="hadoopJob" value="connectedComponentsJob" />
86
						<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
87
						<PARAM name="cluster" ref="cluster"/>
88
						<PARAM name="jobParams">
89
							<MAP>
90
								<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
91
								<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
92
								<ENTRY key="entityType" env="dedup.entity.name"/>
93
								<ENTRY key="entityTypeId" env="dedup.entity.id"/>
94
								<ENTRY key="mapred.input.dir" env="outputPath"/>
95
							</MAP>
96
						</PARAM>
97
					</PARAMETERS>
98
					<ARCS>
99
						<ARC to="markDeleted"/>
100
					</ARCS>
101
				</NODE>
102

    
103
				<NODE name="markDeleted" type="DedupConfigurationLoaderJob">
104
					<DESCRIPTION>mark duplicates as deleted by inference</DESCRIPTION>
105
					<PARAMETERS>
106
						<PARAM name="hadoopJob" value="dedupMarkDeletedEntityJob" />
107
						<PARAM name="dedupConfigSequence" ref="dedupConfigSequence" />
108
						<PARAM name="cluster" ref="cluster"/>
109
						<PARAM name="jobParams">
110
							<MAP>
111
								<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
112
								<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
113
								<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
114
								<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
115
								<ENTRY key="entityType" env="dedup.entity.name"/>
116
								<ENTRY key="entityTypeId" env="dedup.entity.id"/>
117
							</MAP>
118
						</PARAM>
119
					</PARAMETERS>
120
					<ARCS>
121
						<ARC to="buildRoots"/>
122
					</ARCS>
123
				</NODE>
124

    
125

    
126
				<NODE name="buildRoots" type="DedupConfigurationLoaderJob">
127
					<DESCRIPTION>redirect rels</DESCRIPTION>
128
					<PARAMETERS>
129
						<PARAM name="hadoopJob" value="dedupBuildRootsJob" />
130
						<PARAM name="dedupConfigSequence" ref="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 key="entityType" env="dedup.entity.name"/>
139
								<ENTRY key="entityTypeId" env="dedup.entity.id"/>
140
								<ENTRY key="relClasses" env="relClasses"/>
141
							</MAP>
142
						</PARAM>
143
					</PARAMETERS>
144
					<ARCS>
145
						<ARC to="entitySequence"/>
146
					</ARCS>
147
				</NODE>
148

    
149

    
150
			</WORKFLOW>
151
		</CONFIGURATION>
152
	</BODY>
153
</RESOURCE_PROFILE>
(1-1/15)