/ - Diff - D-Net - D-Net project tracking tool

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="6d91b311-a7fd-48ff-98d2-1fed70850e3a_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="coauthorUpdateJob" type="mapreduce">
      			<DESCRIPTION>update coauthors using a map {merged author id --> anchorId}</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.inputformat.class"	value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.CoAuthorUpdateMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.output.compress" value="false" />
     				<PROPERTY key="mapred.compress.map.output" value="true" />
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
             	</JOB_INTERFACE>
             	<SCAN>
                     <FILTERS operator="MUST_PASS_ALL">
                         <FILTER type="prefix" value="30"/>
                     </FILTERS>
                     <FAMILIES>
                         <FAMILY value="person"/>
                     </FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="6363b833-ac88-421c-8596-440a3dc735db_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="deleteSimRelJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that deletes the similarity rels used to in the deduplication process (person)</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupDeleteSimRelMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.Writable" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
             			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
             			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
          				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="073e55eb-c6f4-49a9-80b3-1a927612ba5b_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="offlineHbaseLoad" type="mapreduce">
      			<DESCRIPTION>map reduce job that loads a given entity type in the offline dedup table</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.OfflineHbaseLoadMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.compress.map.output" value="true" />
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="29638605-235b-4cc1-9bf5-a5dd2fc84915_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="dedupPersonJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that scans a given entity type and creates the similarRel graph</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT  -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.SimpleDedupPersonMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<!-- REDUCER -->
     				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.SimpleDedupPersonReducer" />
     				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Writable" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="1000" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="5c4b4dbf-8198-4f7a-9a35-367c7b0a7391_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="dedupSimilarity2ActionsJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that scans a given entity type and creates the similarRel graph</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupSimilarityToActionsMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.compress.map.output" value="true" />
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             		<PARAM name="dedup.conf" required="true" description="dedup configuration" />
             		<PARAM name="rawSetId" required="true" description="raw set identifier" />
             		<PARAM name="similarityCF" required="true" description="similarity column family name" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
                			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
             			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
          				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="1c34963b-75b3-4440-9f42-72445a26c077_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="indexFeedJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that joins the entities on the hbase table and produces a sequence file containig the xml records</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.IndexFeedMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- JOB GLOBAL -->
                     <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
                     <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/>
     				<!-- MISC -->
     				<PROPERTY key="mapred.task.timeout" value="1800000"/>
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<PROPERTY key="mapred.fairscheduler.pool" value="solr"/>
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
             		<PARAM name="mapred.output.dir" required="true" description="destination path on hdfs for rotten index xml records" />
             		<PARAM name="index.solr.url" required="false" description="url used to instantiate the solr client" />
            			<PARAM name="index.solr.collection" required="true" description="target solr collection to be fed" />
            			<PARAM name="id" required="true" description="index DS id" />
     				<PARAM name="index.shutdown.wait.time" required="true" description="wait time before shut down the solr client pool" />
            			<PARAM name="index.buffer.flush.threshold" required="true" description="indexing buffer flush threshold" />
            			<PARAM name="index.feed.timestamp" required="true" description="timestamp used as ds_version" />
     				<PARAM name="index.solr.sim.mode" required="true" description="boolean value, allows to run this job in simulation mode" />
     				<PARAM name="index.xslt" required="true" description="record transformer created by the MSRO service" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS />
             		<FAMILIES />
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="6b2d8db3-346f-4ddc-8591-39fd488c1191_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="dedupGrouperJob" type="mapreduce">
      			<DESCRIPTION>map only job that closes the similarity mesh</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupGrouperMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
             			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
          				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <?xml version="1.0" ?>
     <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/maven-v4_0_0.xsd">
     	<parent>
     		<groupId>eu.dnetlib</groupId>
     		<artifactId>dnet-parent</artifactId>
     		<version>1.0.0</version>
     		<relativePath />
     	</parent>
     	<modelVersion>4.0.0</modelVersion>
     	<groupId>eu.dnetlib</groupId>
     	<artifactId>dnet-openaireplus-profiles</artifactId>
     	<packaging>jar</packaging>
     	<version>1.0.11</version>
     	<scm>
        		<developerConnection>scm:svn:https://svn.driver.research-infrastructures.eu/driver/dnet40/modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.11</developerConnection>
     	</scm>
     	<dependencies>
     	</dependencies>
     </project>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="03d7af20-63bb-4790-a052-6cdbc1e05fce_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2015-02-09T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="oaiFeedJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that feeds the OAI store</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.NullOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.oai.OaiFeedMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.NullWritable" />
     				<!-- JOB GLOBAL -->
     <!--                 <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable" /> -->
     <!--                 <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/> -->
     				<!-- MISC -->
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     				<!--  	Uncomment to override the default lib path -->
     				<PROPERTY key="job.lib" value="/user/dnet/lib/dnet-mapreduce-jobs-assembly-0.0.6.3-SNAPSHOT.jar"/>
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
     				<PARAM name="services.publisher.oai.collection" required="true" description="target mongodb collection" />
     				<PARAM name="oaiConfiguration" required="true" description="configuration bean used to guide the OAI feeding" />
     				<PARAM name="oai.feed.date" required="true" description="timestamp" />
     				<PARAM name="services.publisher.oai.host" required="true" description="mongodb host" />
     				<PARAM name="services.publisher.oai.port" required="true" description="mongodb port" />
     				<PARAM name="services.publisher.oai.db" required="true" description="mongodb database name" />
     				<PARAM name="services.publisher.oai.skipDuplicates" required="true" description="skip duplicated records." />
     				<PARAM name="services.publisher.oai.duplicateXPath" required="true" description="records with this xpath are identified as duplicates" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS />
             		<FAMILIES />
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="ba309300-76f2-40d1-afe3-b77016f443e9_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="exportPersonFullnamesJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that exports the person fullnames on a text file on HDFS</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.preprocess.ExportFullnameMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- REDUCER -->
     				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.preprocess.ExportFullnameReducer" />
     				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="1" />
     				<PROPERTY key="dfs.blocksize" value="256M" />
     				<PROPERTY key="mapred.textoutputformat.separator" value="" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN caching="10">
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES/>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="13beed98-81bf-4fbd-ab4f-de071177997c_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
         	<HADOOP_JOB name="iisPreprocessingQuickJob" type="oozie">
             	<DESCRIPTION>IIS preprocessing</DESCRIPTION>
                 <STATIC_CONFIGURATION>
     				<!-- Cluster wide -->
                     <PROPERTY key="queueName" value="default"/>
     				<PROPERTY key="user.name" value="dnet" />
     				<!-- Runtime -->
                     <PROPERTY key="workingDir" value="/tmp/integration/working_dir/preprocessing_quick_test"/>
                     <PROPERTY key="oozie.wf.validate.ForkJoin" value="false"/>
                     <PROPERTY key="input_referenceextraction_project" value="/user/marek.horst/share/referenceextraction/document_projects/2014-04-11"/>
                     <PROPERTY key="input_referenceextraction_dataset" value="/user/marek.horst/share/referenceextraction/document_datasets/all/2014-04-11"/>
                     <PROPERTY key="export_action_hbase_table_initialize" value="false"/>
                 </STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
      		       	<PARAM name="import_content_object_store_location" required="true" description="mdstore service location" />
             		<PARAM name="import_mdstore_service_location" required="true" description="mdstore service location" />
             		<PARAM name="import_dataset_mdstore_ids_csv" required="true" description="mdstore id for dataset records" />
             		<PARAM name="import_wos_mdstore_id" required="true" description="mdstore id for WoS records" />
             		<PARAM name="import_database_service_location" required="true" description="database service endpoint" />
             		<PARAM name="import_content_datacite_objectstores_csv" required="true" description="objecstore ids subject to dataset reference extraction" />
             		<PARAM name="import_content_wos_plaintext_objectstores_csv" required="true" description="objecstore ids for WoS plaintext" />
     	      		<PARAM name="export_action_hbase_table_name" required="true" description="destination action manager table" />
             		<PARAM name="export_action_hbase_remote_zookeeper_quorum" required="true" description="ZK quorum" />
             		<PARAM name="export_action_hbase_remote_zookeeper_clientport" required="true" description="ZK port" />
             		<PARAM name="nameNode" required="true" description="hdfs name node" />
             		<PARAM name="jobTracker" required="true" description="job tracker name" />
             		<PARAM name="oozie.wf.application.path" required="true" description="oozie job application absolute path" />
            			<PARAM name="export_action_set_id_document_referencedProjects" required="true" description="target action set for project references" />
            			<PARAM name="export_action_set_id_document_referencedDatasets" required="true" description="target action set for dataset references" />
            			<PARAM name="export_action_set_id_entity_wos" required="true" description="target action set for WoS entities" />
          			<PARAM name="export_action_set_id_entity_dataset" required="true" description="target action set for dataset entities" />
             	</JOB_INTERFACE>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="4c63a9ab-057f-442c-8da2-9b956c41e645_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="dedupRootsPersonExportJob" type="mapreduce">
      			<DESCRIPTION>map only job that exports the representative publications as json</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.RootPersonExportMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="dfs.blocksize" value="256M" />
     				<PROPERTY key="mapred.reduce.tasks" value="1" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
     	       			<FAMILY param="entityType" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="bc4f377a-af07-403d-a019-af60aa557652_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="resetDedupJob" type="mapreduce">
      			<DESCRIPTION>map only job that reset the dedup jobs</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"  />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
     				<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.reset.HBaseResetMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"  />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Mutation"  />
     				<!-- MISC -->
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"  />
     				<PROPERTY key="mapreduce.map.speculative" value="false"  />
     				<PROPERTY key="mapred.reduce.tasks" value="0"  />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="888ef72f-701a-4d59-8b8a-2ad01986f975_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="gtCleanerJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that deletes the non-GT rows</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.GTCleanerMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Delete" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
             			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
             			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
          				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="719b5d2b-4156-4936-bbc3-41d908ec3c57_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="elastisearchTestJob" type="mapreduce">
      			<DESCRIPTION>map only job that indexes over ES</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.elasticsearch.hadoop.mr.EsOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.es.ElasticsearchFeedMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.BytesWritable" />
     				<!-- MISC -->
     				<PROPERTY key="es.nodes" value="146.48.87.110:9200" />
     				<PROPERTY key="es.resource" value="openaire/oaf" />
     				<PROPERTY key="es.input.json" value="yes" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS />
             		<FAMILIES />
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="027554bd-3d5c-4c50-9170-90d8c4402bc3_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="dedupRootsToCSVJob" type="mapreduce">
      			<DESCRIPTION>map only job that exports the representatives as CSV files</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.LazyOutputFormat" />
     				<PROPERTY key="mapreduce.output.lazyoutputformat.outputformat" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupRootsToCsvMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<!-- REDUCER -->
     				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupRootsToCsvReducer" />
     				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- MUTIPLE OUTPUT -->
     				<PROPERTY key="mapreduce.multipleoutputs" value="NativeGroups Groups NativeEntities" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.key" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.value" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.key" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.value" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.key" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.value" value="org.apache.hadoop.io.Text" />
     				<!-- MISC -->
                     <PROPERTY key="mapred.textoutputformat.wrapper" value="#"/>
                     <PROPERTY key="mapred.textoutputformat.separator" value="!"/>
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="3" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
     	       			<FAMILY param="entityType" />
             			<FAMILY value="resultResult_dedup_merges" />
             			<FAMILY value="personPerson_dedup_merges" />
          				<FAMILY value="organizationOrganization_dedup_merges" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="c611ec67-eefc-4ffe-a5d4-cb3fc40a8baf_RGVkdXBDb25maWd1cmF0aW9uRFNSZXNvdXJjZXMvRGVkdXBDb25maWd1cmF0aW9uRFNSZXNvdXJjZVR5cGU="/>
             <RESOURCE_TYPE value="DedupConfigurationDSResourceType"/>
             <RESOURCE_KIND value="DedupConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <CONFIGURATION>
     	        <DESCRIPTION>1 - Publication: Match against the title, whose numbers must match</DESCRIPTION>
                 <DEDUPLICATION>
+     			{
     				"wf" : {
     			        "threshold" : "0.99",
     			        "dedupRun" : "001",
     			        "entityType" : "result",
     			        "orderField" : "title",
     			        "queueMaxSize" : "2000",
     			        "groupMaxSize" : "10",
     			        "slidingWindowSize" : "200",
     			        "rootBuilder" : [ "result", "personResult_authorship_hasAuthor", "resultProject_outcome_isProducedBy", "resultResult_publicationDataset_isRelatedTo", "resultResult_similarity_isAmongTopNSimilarDocuments", "resultResult_similarity_hasAmongTopNSimilarDocuments" ],
     			        "includeChildren" : "true"
     			    },
     				"pace" : {
     					"clustering" : [
     						{ "name" : "ngrampairs", "fields" : [ "title" ], "params" : { "max" : "1", "ngramLen" : "3"} },
     						{ "name" : "suffixprefix", "fields" : [ "title" ], "params" : { "max" : "1", "len" : "3" } }
     					],
     			  		"conditions" : [
     			  			{ "name" : "titleVersionMatch", "fields" : [ "title" ] },
     			  			{ "name" : "sizeMatch", "fields" : [ "authors" ] }
     			  		],
     					"model" : [
     						{ "name" : "title", "algo" : "LevensteinTitle", "type" : "String", "weight" : "1.0", "ignoreMissing" : "false", "path" : "result/metadata/title/value" },
     						{ "name" : "authors", "algo" : "Null", "type" : "List", "weight" : "0.0", "ignoreMissing" : "true", "path" : "result/author/metadata/fullname/value" }
     					],
     					"blacklists" : {
     						"title" : [
     				"^THE ASSOCIATION AND THE GENERAL MEDICAL COUNCIL$",
     				"^Problems with perinatal pathology\.?$",
     				"(?i)^Cases? of Puerperal Convulsions$",
     				"(?i)^Operative Gyna?ecology$",
     				"(?i)^Mind the gap\!?\:?$",
     				"^Chronic fatigue syndrome\.?$",
     				"^Cartas? ao editor Letters? to the Editor$",
     				"^Note from the Editor$",
     				"^Anesthesia Abstract$",
     				"^Annual report$",
     				"(?i)^“?THE RADICAL PREVENTION OF VENEREAL DISEASE\.?”?$",
     				"(?i)^Graph and Table of Infectious Diseases?$",
     				"^Presentation$",
     				"(?i)^Reviews and Information on Publications$",
     				"(?i)^PUBLIC HEALTH SERVICES?$",
     				"(?i)^COMBINED TEXT-?BOOK OF OBSTETRICS AND GYN(Æ|ae)COLOGY$",
     				"(?i)^Adrese autora$",
     				"(?i)^Systematic Part .*\. Catalogus Fossilium Austriae, Band 2: Echinoidea neogenica$",
     				"(?i)^Acknowledgement to Referees$",
     				"(?i)^Behçet's disease\.?$",
     				"(?i)^Isolation and identification of restriction endonuclease.*$",
     				"(?i)^CEREBROVASCULAR DISEASES?.?$",
     				"(?i)^Screening for abdominal aortic aneurysms?\.?$",
     				"^Event management$",
     				"(?i)^Breakfast and Crohn's disease.*\.?$",
     				"^Cálculo de concentraciones en disoluciones acuosas. Ejercicio interactivo\..*\.$",
     				"(?i)^Genetic and functional analyses of SHANK2 mutations suggest a multiple hit model of Autism spectrum disorders?\.?$",
     				"^Gushi hakubutsugaku$",
     				"^Starobosanski nadpisi u Bosni i Hercegovini \(.*\)$",
     				"^Intestinal spirocha?etosis$",
     				"^Treatment of Rodent Ulcer$",
     				"(?i)^\W*Cloud Computing\W*$",
     				"^Compendio mathematico : en que se contienen todas las materias mas principales de las Ciencias que tratan de la cantidad$",
     				"^Free Communications, Poster Presentations: Session [A-F]$",
     				"^“The Historical Aspects? of Quackery\.?”$",
     				"^A designated centre for people with disabilities operated by St John of God Community Services (Limited|Ltd), Louth$",
     				"^P(er|re)-Mile Premiums for Auto Insurance\\.?$",
     				"(?i)^Case Report$",
     				"^Boletín Informativo$",
     				"(?i)^Glioblastoma Multiforme$",
     				"(?i)^Nuevos táxones animales descritos en la península Ibérica y Macaronesia desde 1994 \\(.*\\)$",
     				"^Zaměstnanecké výhody$",
     				"(?i)^The Economics of Terrorism and Counter-Terrorism: A Survey \\(Part .*\\)$",
     				"(?i)^Carotid body tumours?\\.?$",
     				"(?i)^\\[Españoles en Francia : La condición Emigrante.*\\]$",
     				"^Avant-propos$",
     				"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Head(s)? and Capital(s)?$",
     				"(?i)^St\. Patrick's Cathedral, Dublin, County Dublin - Bases?$",
     				"(?i)^PUBLIC HEALTH VERSUS THE STATE$",
     				"^Viñetas de Cortázar$",
     				"(?i)^Search for heavy neutrinos and W(\\[|_|\\(|_\\{|-)?R(\\]|\\)|\\})? bosons with right-handed couplings in a left-right symmetric model in pp collisions at.*TeV(\.)?$",
     				"(?i)^Measurement of the pseudorapidity and centrality dependence of the transverse energy density in Pb(-?)Pb collisions at.*tev(\.?)$",
     				"(?i)^Search for resonances decaying into top-quark pairs using fully hadronic decays in pp collisions with ATLAS at.*TeV$",
     				"(?i)^Search for neutral minimal supersymmetric standard model Higgs bosons decaying to tau pairs in pp collisions at.*tev$",
     				"(?i)^Relatório de Estágio (de|em) Angiologia e Cirurgia Vascular$",
     				"^Aus der AGMB$",
     				"^Znanstveno-stručni prilozi$",
     				"^Zhodnocení finanční situace podniku a návrhy na zlepšení$",
     				"^Evaluation of the Financial Situation in the Firm and Proposals to its Improvement$",
     				"^Finanční analýza podniku$",
     				"^Financial analysis( of business)?$",
     				"(?i)^Textbook of Gyn(a)?(Æ)?(e)?cology$",
     				"^Jikken nihon shūshinsho$",
     				"(?i)^CORONER('|s)(s|') INQUESTS$",
     				"(?i)^(Μελέτη παραγόντων )?risk management( για ανάπτυξη και εφαρμογή ενός πληροφοριακού συστήματος| και ανάπτυξη συστήματος)?$",
     				"(?i)^Consultants' contract(s)?$",
     				"(?i)^Upute autorima$",
     				"(?i)^Bijdrage tot de Kennis van den Godsdienst der Dajaks van Lan(d|f)ak en Tajan$",
     				"^Joshi shin kokubun$",
     				"^Kōtō shōgaku dokuhon nōson'yō$",
     				"^Jinjō shōgaku shōka$",
     				"^Shōgaku shūjichō$",
     				"^Nihon joshi dokuhon$",
     				"^Joshi shin dokuhon$",
     				"^Chūtō kanbun dokuhon$",
     				"^Wabun dokuhon$",
     				"(?i)^(Analysis of economy selected village or town|Rozbor hospodaření vybrané obce či města)$",
     				"(?i)^cardiac rehabilitation$",
     				"(?i)^Analytical summary$",
     				"^Thesaurus resolutionum Sacrae Congregationis Concilii$",
     				"(?i)^Sumario analítico(\\s{1})?(Analitic summary)?$",
     				"^Prikazi i osvrti$",
     				"^Rodinný dům s provozovnou$",
     				"^Family house with an establishment$",
     				"^Shinsei chūtō shin kokugun$",
     				"^Pulmonary alveolar proteinosis(\\.?)$",
     				"^Shinshū kanbun$",
     				"^Viñeta(s?) de Rodríguez$",
     				"(?i)^RUBRIKA UREDNIKA$",
     				"^A Matching Model of the Academic Publication Market$",
     				"^Yōgaku kōyō$",
     				"^Internetový marketing$",
     				"^Internet marketing$",
     				"^Chūtō kokugo dokuhon$",
     				"^Kokugo dokuhon$",
     				"^Antibiotic Cover for Dental Extraction(s?)$",
     				"^Strategie podniku$",
     				"^Strategy of an Enterprise$",
     				"(?i)^respiratory disease(s?)(\.?)$",
     				"^Award(s?) for Gallantry in Civil Defence$",
     				"^Podniková kultura$",
     				"^Corporate Culture$",
     				"^Severe hyponatraemia in hospital inpatient(s?)(\.?)$",
     				"^Pracovní motivace$",
     				"^Work Motivation$",
     				"^Kaitei kōtō jogaku dokuhon$",
     				"^Konsolidovaná účetní závěrka$",
     				"^Consolidated Financial Statements$",
     				"(?i)^intracranial tumour(s?)$",
     				"^Climate Change Mitigation Options and Directed Technical Change: A Decentralized Equilibrium Analysis$",
     				"^\\[CERVECERIAS MAHOU(\\.|\\:) INTERIOR\\] \\[Material gráfico\\]$",
     				"^Housing Market Dynamics(\\:|\\.) On the Contribution of Income Shocks and Credit Constraint(s?)$",
     				"^\\[Funciones auxiliares de la música en Radio París,.*\\]$",
     				"^Úroveň motivačního procesu jako způsobu vedení lidí$",
     				"^The level of motivation process as a leadership$",
     				"^Pay-beds in N(\\.?)H(\\.?)S(\\.?) Hospitals$",
     				"(?i)^news and events$",
     				"(?i)^NOVOSTI I DOGAĐAJI$",
     				"^Sansū no gakushū$",
     				"^Posouzení informačního systému firmy a návrh změn$",
     				"^Information System Assessment and Proposal for ICT Modification$",
     				"^Stresové zatížení pracovníků ve vybrané profesi$",
     				"^Stress load in a specific job$",
     				"^Sunday: Poster Sessions, Pt.*$",
     				"^Monday: Poster Sessions, Pt.*$",
     				"^Wednesday: Poster Sessions, Pt.*",
     				"^Tuesday: Poster Sessions, Pt.*$",
     				"^Analýza reklamy$",
     				"^Analysis of advertising$",
     				"^Shōgaku shūshinsho$",
     				"^Shōgaku sansū$",
     				"^Shintei joshi kokubun$",
     				"^Taishō joshi kokubun dokuhon$",
     				"^Joshi kokubun$",
     				"^Účetní uzávěrka a účetní závěrka v ČR$",
     				"(?i)^The \"?Causes\"? of Cancer$",
     				"^Normas para la publicación de artículos$",
     				"^Editor('|s)(s|') [Rr]eply$",
     				"^Editor(’|s)(s|’) letter$",
     				"^Redaktoriaus žodis$",
     				"^DISCUSSION ON THE PRECEDING PAPER$",
     				"^Kōtō shōgaku shūshinsho jidōyō$",
     				"^Shōgaku nihon rekishi$",
     				"^(Theory of the flow of action currents in isolated myelinated nerve fibers).*$",
     				"^Préface$",
     				"^Occupational [Hh]ealth [Ss]ervices.$",
     				"^In Memoriam Professor Toshiyuki TAKESHIMA$",
     				"^Účetní závěrka ve vybraném podniku.*$",
     				"^Financial statements in selected company$",
     				"^Abdominal [Aa]ortic [Aa]neurysms.*$",
     				"^Pseudomyxoma peritonei$",
     				"^Kazalo autora$",
     				"(?i)^uvodna riječ$",
     				"^Motivace jako způsob vedení lidí$",
     				"^Motivation as a leadership$",
     				"^Polyfunkční dům$",
     				"^Multi\\-funkcional building$",
     				"^Podnikatelský plán$",
     				"^Business Plan$",
     				"^Oceňování nemovitostí$",
     				"^Marketingová komunikace$",
     				"^Marketing communication$",
     				"^Sumario Analítico$",
     				"^Riječ uredništva$",
     				"^Savjetovanja i priredbe$",
     				"^Índice$",
     				"^(Starobosanski nadpisi).*$",
     				"^Vzdělávání pracovníků v organizaci$",
     				"^Staff training in organization$",
     				"^(Life Histories of North American Geometridae).*$",
     				"^Strategická analýza podniku$",
     				"^Strategic Analysis of an Enterprise$",
     				"^Sadržaj$",
     				"^Upute suradnicima$",
     				"^Rodinný dům$",
     				"(?i)^Fami(l)?ly house$",
     				"^Upute autorima$",
     				"^Strategic Analysis$",
     				"^Finanční analýza vybraného podniku$",
     				"^Finanční analýza$",
     				"^Riječ urednika$",
     				"(?i)^Content(s?)$",
     				"(?i)^Inhalt$",
     				"^Jinjō shōgaku shūshinsho jidōyō$",
     				"(?i)^Index$",
     				"^Chūgaku kokubun kyōkasho$",
     				"^Retrato de una mujer$",
     				"^Retrato de un hombre$",
     				"^Kōtō shōgaku dokuhon$",
     				"^Shotōka kokugo$",
     				"^Shōgaku dokuhon$",
     				"^Jinjō shōgaku kokugo dokuhon$",
     				"^Shinsei kokugo dokuhon$",
     				"^Teikoku dokuhon$",
     				"^Instructions to Authors$",
     				"^KİTAP TAHLİLİ$",
     				"^PRZEGLĄD PIŚMIENNICTWA$",
     				"(?i)^Presentación$",
     				"^İçindekiler$",
     				"(?i)^Tabl?e of contents$",
     				"^(CODICE DEL BEATO DE LOS REYES FERNANDO I Y SANCHA).*$",
     				"^(\\[MADRID\\. BIBL\\. NAC\\. N.*KING FERDINAND I.*FROM SAN ISIDORO DE LEON\\. FACUNDUS SCRIPSIT DATED.*\\]).*",
     				"^Editorial( Board)?$",
     				"(?i)^Editorial \\(English\\)$",
     				"^Editörden$",
     				"^(Corpus Oral Dialectal \\(COD\\)\\.).*$",
     				"^(Kiri Karl Morgensternile).*$",
     				"^(\\[Eksliibris Aleksandr).*\\]$",
     				"^(\\[Eksliibris Aleksandr).*$",
     				"^(Eksliibris Aleksandr).*$",
     				"^(Kiri A\\. de Vignolles).*$",
     				"^(2 kirja Karl Morgensternile).*$",
     				"^(Pirita kloostri idaosa arheoloogilised).*$",
     				"^(Kiri tundmatule).*$",
     				"^(Kiri Jenaer Allgemeine Literaturzeitung toimetusele).*$",
     				"^(Eksliibris Nikolai Birukovile).*$",
     				"^(Eksliibris Nikolai Issakovile).*$",
     				"^(WHP Cruise Summary Information of section).*$",
     				"^(Measurement of the top quark\\-pair production cross section with ATLAS in pp collisions at).*$",
     				"^(Measurement of the spin\\-dependent structure function).*"
+    						]
+    					}
+    				}
+    			}
                 </DEDUPLICATION>
             </CONFIGURATION>
             <STATUS>
                 <LAST_UPDATE value="2001-12-31T12:00:00"/>
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="5626c94e-0005-416a-9ea4-48fc8af85ecd_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
             <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
             <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <HADOOP_JOB name="deleteDedupRelsJob" type="mapreduce">
      			<DESCRIPTION>map reduce job that deletes the dedup rels used to in the deduplication process</DESCRIPTION>
             	<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
             		<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupDeleteRelMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.Writable" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="0" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     		<!--  	Uncomment to override the default lib path -->
     		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
             	</STATIC_CONFIGURATION>
             	<JOB_INTERFACE>
             		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
             		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
             		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
             	</JOB_INTERFACE>
             	<SCAN>
             		<FILTERS operator="MUST_PASS_ALL">
             			<FILTER type="prefix" param="entityTypeId" />
             		</FILTERS>
             		<FAMILIES>
             			<FAMILY param="entityType" />
     			        <FAMILY value="resultResult_dedup_merges" />
     			        <FAMILY value="resultResult_dedup_isMergedIn" />
     			        <FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
     			        <FAMILY value="personPerson_dedup_merges" />
     			        <FAMILY value="personPerson_dedup_isMergedIn" />
     			        <FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
     			        <FAMILY value="organizationOrganization_dedup_merges" />
     			        <FAMILY value="organizationOrganization_dedup_isMergedIn" />
     			        <FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
             		</FAMILIES>
             	</SCAN>
             </HADOOP_JOB>
             <STATUS>
                 <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
                 <RUNNING_INSTANCES value="0"/>
                 <CUMULATIVE_RUN value="0" />
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
         <HEADER>
             <RESOURCE_IDENTIFIER value="82b1c7fb-c36c-4291-8863-0393c7c588ee_RGVkdXBDb25maWd1cmF0aW9uRFNSZXNvdXJjZXMvRGVkdXBDb25maWd1cmF0aW9uRFNSZXNvdXJjZVR5cGU="/>
             <RESOURCE_TYPE value="DedupConfigurationDSResourceType"/>
             <RESOURCE_KIND value="DedupConfigurationDSResources"/>
             <RESOURCE_URI value=""/>
             <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
         </HEADER>
         <BODY>
             <CONFIGURATION>
                	<DESCRIPTION>1 - Person: Decision tree</DESCRIPTION>
                 <DEDUPLICATION>
+    			{
     				"wf" : {
     			        "threshold" : "1.0",
     			        "dedupRun" : "001",
     			        "entityType" : "person",
     			        "orderField" : "fullname",
     			        "queueMaxSize" : "2000",
     			        "groupMaxSize" : "10",
     			        "slidingWindowSize" : "200",
     			        "rootBuilder" : [ "person" ],
     			        "includeChildren" : "true"
     			    },
     				"pace" : {
     					"clustering" : [
     						{ "name" : "personclustering", "fields" : [ "person" ], "params" : { } }
     					],
     					"model" : [
     			            { "name" : "fullname", "algo" : "JaroWinkler", "type" : "String", "weight" : "0.3", "ignoreMissing" : "false", "path" : "person/metadata/fullname/value", "params" : { } },
     			            { "name" : "person", "algo" : "PersonDistance", "type" : "JSON", "weight" : "0.7", "ignoreMissing" : "false", "path" : "person", "params" : { "common.anchors" : "1", "common.surnames" : "3" } },
     			            { "name" : "lastname", "algo" : "Null", "type" : "String", "weight" : "0", "ignoreMissing" : "true", "path" : "person/metadata/secondnames/value" }
     					],
     					"blacklists" : {
     			            "lastname" : [
     				            "(?i)^wang$",
     				            "(?i)^~wang$",
     				            "(?i)^zhang$",
     				            "(?i)^zhou$",
     				            "(?i)^zhao$",
     				            "(?i)^li$",
     				            "(?i)^~li$",
     				            "(?i)^liu$",
     				            "(?i)^chen$",
     				            "(?i)^yang$",
     				            "(?i)^kim$",
     				            "(?i)^xu$",
     				            "(?i)^huang$",
     				            "(?i)^sun$",
     				            "(?i)^lee$",
     				            "(?i)^ma$",
     				            "(?i)^kim$",
     				            "(?i)^hu$",
     				            "(?i)^wu$",
     				            "(?i)^zhu$",
     				            "(?i)^lu$"
+    			            ]
+    	                }
+    				}
+    			}
                 </DEDUPLICATION>
             </CONFIGURATION>
             <STATUS>
                 <LAST_UPDATE value="2001-12-31T12:00:00"/>
             </STATUS>
             <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
         </BODY>
     </RESOURCE_PROFILE>

     <RESOURCE_PROFILE>
     	<HEADER>
     		<RESOURCE_IDENTIFIER value="3f544a36-f123-4f5c-acf4-7c25f6591ec4_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
     		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
     		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
     		<RESOURCE_URI value=""/>
     		<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
     	</HEADER>
     	<BODY>
     		<HADOOP_JOB name="personCsvJoinJob" type="mapreduce">
     			<DESCRIPTION>map reduce job that joins person entities by "surname+first name letter" and serialise the output as csv</DESCRIPTION>
     			<STATIC_CONFIGURATION>
     				<!-- I/O FORMAT -->
     				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
     				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.LazyOutputFormat" />
     				<PROPERTY key="mapreduce.output.lazyoutputformat.outputformat" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
     				<!-- MAPPER -->
     				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.JoinPersonGroupMapper" />
     				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- REDUCER -->
     				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.JoinPersonGroupReducer" />
     				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
     				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
     				<!-- MISC -->
     				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
     				<PROPERTY key="mapreduce.map.speculative" value="false" />
     				<PROPERTY key="mapred.reduce.tasks" value="10" />
     				<!-- <PROPERTY key="user.name" value="dnet" /> -->
     				<!--  	Uncomment to override the default lib path -->
     				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
     			</STATIC_CONFIGURATION>
     			<JOB_INTERFACE>
     				<PARAM name="mapred.input.dir" required="true" description="input sequence file" />
     			</JOB_INTERFACE>
     		</HADOOP_JOB>
     		<STATUS>
     			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
     			<RUNNING_INSTANCES value="0"/>
     			<CUMULATIVE_RUN value="0" />
     		</STATUS>
     		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
     	</BODY>

Project

General

Profile

D-Net

Revision 42411

Added by Claudio Atzori over 8 years ago