2022-01-03 14:33:00 +01:00
<RESOURCE_PROFILE >
<HEADER >
<RESOURCE_IDENTIFIER value= "4801c33c-66ca-4ab6-af64-aa812194ec61_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" />
<RESOURCE_TYPE value= "WorkflowDSResourceType" />
<RESOURCE_KIND value= "WorkflowDSResources" />
<RESOURCE_URI value= "" />
2022-01-21 13:59:46 +01:00
<DATE_OF_CREATION value= "2022-01-11T20:04:48+00:00" />
2022-01-03 14:33:00 +01:00
</HEADER>
<BODY >
<WORKFLOW_NAME > Graph processing [EXPERIMENT]</WORKFLOW_NAME>
<WORKFLOW_TYPE > Data Provision</WORKFLOW_TYPE>
<WORKFLOW_PRIORITY > 30</WORKFLOW_PRIORITY>
<CONFIGURATION start= "manual" >
<NODE isStart= "true" name= "setUnresolvedEntityPath" type= "SetEnvParameter" >
<DESCRIPTION > set the path of unresolved entities</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > unresolvedEntityPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /data/unresolved_BETA/content</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setNsPrefixBlacklistForBETA" type= "SetEnvParameter" >
<DESCRIPTION > set blacklist of funder nsPrefixes from the beta aggregator</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > nsPrefixBlacklist_BETA</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > gsrt________,rcuk________</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setNsPrefixBlacklistForPROD" type= "SetEnvParameter" >
<DESCRIPTION > set blacklist of funder nsPrefixes from the production aggregator</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > nsPrefixBlacklist_PROD</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > gsrt________,rcuk________,fct_________,nwo_________</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setIdMappingPath" type= "SetEnvParameter" >
<DESCRIPTION > set the path of the map defining the relations id mappings</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > idMappingPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /data/maps/fct_map.json</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
2022-01-21 13:59:46 +01:00
<NODE isStart= "true" name= "setIterationNumbers" type= "SetEnvParameter" >
<DESCRIPTION > set the number of iteration in affiliation propagation</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > iterations</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > 1</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
2022-01-03 14:33:00 +01:00
<NODE isStart= "true" name= "setMergedGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the MERGED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > mergedGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/01_graph_merged</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setRawGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the RAW graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > rawGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/02_graph_raw</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setCleanedFirstGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the the consistent graph cleaned</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > cleanedFirstGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/03_graph_cleaned</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setResolvedGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the CLEANED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > resolvedGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/04_graph_resolved</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
2022-01-21 13:59:46 +01:00
<NODE isStart= "true" name= "setGroupedGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the GROUPED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > groupedGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/05_graph_grouped</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
2022-01-03 14:33:00 +01:00
<NODE isStart= "true" name= "setInferredGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the INFERRED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > inferredGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/06_graph_inferred</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setDedupGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the DEDUPED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > dedupGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/07_graph_dedup</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setConsistentGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the CONSISTENCY graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > consistentGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/08_graph_consistent</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setOrcidGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the ORCID enriched graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > orcidGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/09_graph_orcid</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setBulkTaggingGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the BULK TAGGED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > bulkTaggingGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/10_graph_bulktagging</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setAffiliationGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the AFFILIATION from INSTITUTIONAL REPOS graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > affiliationGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/11_graph_affiliation</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setAffiliationSemRelGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the AFFILIATION from SEMATIC RELATION graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > affiliationSemRelGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/12_graph_affiliationsr</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setCommunityOrganizationGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the COMMUNITY from SELECTED SOURCES graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > communityOrganizationGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/13_graph_community_organization</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setFundingGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the FUNDING from SEMANTIC RELATION graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > fundingGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/14_graph_funding</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setCommunitySemRelGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the COMMUNITY from SEMANTIC RELATION graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > communitySemRelGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/15_graph_community_sem_rel</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setCountryGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the COUNTRY enriched graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > countryGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/16_graph_country</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setCleanedGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the CLEANED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > cleanedGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/17_graph_cleaned</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setBlacklistedGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the target path to store the blacklisted graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > blacklistedGraphPath</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/18_graph_blacklisted</PARAM>
2022-01-03 14:33:00 +01:00
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setBulkTaggingPathMap" type= "SetEnvParameter" >
<DESCRIPTION > Set the map of paths for the Bulk Tagging</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > bulkTaggingPathMap</PARAM>
<PARAM managedBy= "system" name= "parameterValue" required= "true" type= "string" > {"author" : "$['author'][*]['fullname']", "title" : "$['title'][*]['value']", "orcid" : "$['author'][*]['pid'][*][?(@['key']=='ORCID')]['value']", "contributor" : "$['contributor'][*]['value']", "description" : "$['description'][*]['value']"}</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setPropagationOrganizationCommunityMap" type= "SetEnvParameter" >
<DESCRIPTION > Set the map of associations organization, community list for the propagation of community to result through organization</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > propagationOrganizationCommunityMap</PARAM>
<PARAM managedBy= "system" name= "parameterValue" required= "true" type= "string" > {"20|corda__h2020::3fb05a9524c3f790391261347852f638":["mes","euromarine"], "20|corda__h2020::e8dbe14cca9bf6fce09d468872f813f8":["mes","euromarine"], "20|snsf________::9b253f265e3bef5cae6d881fdf61aceb":["mes","euromarine"],"20|rcuk________::e054eea0a47665af8c3656b5785ccf76":["mes","euromarine"],"20|corda__h2020::edc18d67c9b11fb616ca9f6e1db1b151":["mes","euromarine"],"20|rcuk________::d5736d9da90521ddcdc7828a05a85e9a":["mes","euromarine"],"20|corda__h2020::f5d418d3aa1cf817ddefcc3fdc039f27":["mes","euromarine"],"20|snsf________::8fa091f8f25a846779acb4ea97b50aef":["mes","euromarine"],"20|corda__h2020::81e020977211c2c40fae2e1a50bffd71":["mes","euromarine"],"20|corda_______::81e020977211c2c40fae2e1a50bffd71":["mes","euromarine"],"20|snsf________::31d0a100e54e3cdb3c6f52d91e638c78":["mes","euromarine"],"20|corda__h2020::ea379ef91b8cc86f9ac5edc4169292db":["mes","euromarine"],"20|corda__h2020::f75ee2ee48e5cb0ec8c8d30aaa8fef70":["mes","euromarine"],"20|rcuk________::e16010089551a1a9182a94604fc0ea59":["mes","euromarine"],"20|corda__h2020::38531a2cce7c5c347ffc439b07c1f43b":["mes","euromarine"],"20|corda_______::38531a2cce7c5c347ffc439b07c1f43b":["mes","euromarine"],"20|grid________::b2cbbf5eadbbf87d534b022bad3191d7":["mes","euromarine"],"20|snsf________::74730ef1439d7f7636a8be58a6b471b8":["mes","euromarine"],"20|nsf_________::ad72e19043a5a467e35f9b444d11563e":["mes","euromarine"],"20|rcuk________::0fc3e92500290902a2d38ec2445e74c3":["mes","euromarine"],"20|grid________::ad2c29905da0eb3c06b3fa80cacd89ea":["mes","euromarine"],"20|corda__h2020::30b53e4d63d3724f00acb9cbaca40860":["mes","euromarine"],"20|corda__h2020::f60f84bee14ad93f0db0e49af1d5c317":["mes","euromarine"], "20|corda__h2020::7bf251ac3765b5e89d82270a1763d09f":["mes","euromarine"], "20|corda__h2020::65531bd11be9935948c7f2f4db1c1832":["mes","euromarine"], "20|corda__h2020::e0e98f86bbc76638bbb72a8fe2302946":["mes","euromarine"], "20|snsf________::3eb43582ac27601459a8d8b3e195724b":["mes","euromarine"], "20|corda__h2020::af2481dab65d06c8ea0ae02b5517b9b6":["mes","euromarine"], "20|corda__h2020::c19d05cfde69a50d3ebc89bd0ee49929":["mes","euromarine"], "20|corda__h2020::af0bfd9fc09f80d9488f56d71a9832f0":["mes","euromarine"], "20|rcuk________::f33c02afb0dc66c49d0ed97ca5dd5cb0":["beopen"],
"20|grid________::a867f78acdc5041b34acfe4f9a349157":["beopen"], "20|grid________::7bb116a1a9f95ab812bf9d2dea2be1ff":["beopen"], "20|corda__h2020::6ab0e0739dbe625b99a2ae45842164ad":["beopen"], "20|corda__h2020::8ba50792bc5f4d51d79fca47d860c602":["beopen"], "20|corda_______::8ba50792bc5f4d51d79fca47d860c602":["beopen"], "20|corda__h2020::e70e9114979e963eef24666657b807c3":["beopen"], "20|corda_______::e70e9114979e963eef24666657b807c3":["beopen"], "20|corda_______::15911e01e9744d57205825d77c218737":["beopen"], "20|opendoar____::056a41e24e2a9a67215e87bbee6a80ab":["beopen"], "20|opendoar____::7f67f2e6c6fbb0628f8160fcd3d92ae3":["beopen"], "20|grid________::a8ecfd7c084e561168bcbe6bf0daf3e3":["beopen"], "20|corda_______::7bbe6cc5d8ec1864739a04b0d020c9e9":["beopen"], "20|corda_______::3ff558e30c2e434d688539548300b050":["beopen"], "20|corda__h2020::5ffee5b3b83b33a8cf0e046877bd3a39":["beopen"], "20|corda__h2020::5187217e2e806a6df3579c46f82401bc":["beopen"], "20|grid________::5fa7e2709bcd945e26bfa18689adeec1":["beopen"], "20|corda_______::d8696683c53027438031a96ad27c3c07":["beopen"], "20|corda__h2020::d8696683c53027438031a96ad27c3c07":["beopen"], "20|rcuk________::23a79ebdfa59790864e4a485881568c1":["beopen"], "20|corda__h2020::b76cf8fe49590a966953c37e18608af9":["beopen"], "20|grid________::d2f0204126ee709244a488a4cd3b91c2":["beopen"], "20|corda__h2020::05aba9d2ed17533d15221e5655ac11e6":["beopen"], "20|grid________::802401579481dc32062bdee69f5e6a34":["beopen"], "20|corda__h2020::3f6d9d54cac975a517ba6b252c81582d":["beopen"]}
</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setDedupConfig" type= "SetEnvParameter" >
<DESCRIPTION > Set the dedup orchestrator name</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > dedupConfig</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > dedup-similarity-result-decisiontree-v2</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "actionSetsRaw" type= "SetEnvParameter" >
<DESCRIPTION > declares the ActionSet ids to promote in the RAW graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > actionSetIdsRawGraph</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > doiboost,orcidworks-no-doi,opencitations,h2020classification</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "actionSetsIIS" type= "SetEnvParameter" >
<DESCRIPTION > declares the ActionSet ids to promote in the INFERRED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > actionSetIdsIISGraph</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > iis-researchinitiative,iis-document-citations,iis-document-citations-relations,iis-document-affiliation,iis-document-classes,iis-document-similarities,iis-referenced-datasets-main,iis-referenced-projects-main,iis-referenceextraction-pdb,document_software_url,iis-extracted-metadata,iis-communities,iis-referenced-patents,iis-covid-19,iis-entities-software,iis-entities-patent</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isStart= "true" name= "setIsLookUpUrl" type= "SetEnvParameter" >
<DESCRIPTION > Set the IS lookup service address</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > isLookUpUrl</PARAM>
<PARAM managedBy= "system" name= "parameterValue" required= "true" type= "string" > http://beta.services.openaire.eu:8280/is/services/isLookUp?wsdl</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig" />
</ARCS>
</NODE>
<NODE isJoin= "true" name= "waitConfig" >
<DESCRIPTION > wait configurations</DESCRIPTION>
<PARAMETERS />
<ARCS >
<ARC to= "reuseODFClaims_PROD" />
<ARC to= "reuseODFClaims_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseODFClaims_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF claims from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODFClaims_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseODF_hdfs_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseODF_hdfs_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF records on HDFS from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODFhdfs_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAFClaims_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseOAFClaims_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF claims from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAFClaims_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAF_hdfs_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseOAF_hdfs_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF records on HDFS from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAFhdfs_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseDB_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseDB_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached DB content from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseDB_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseDBOpenorgs_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseDBOpenorgs_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OpenOrgs content from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseDBOpenorgs_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseODF_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseODF_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF content from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODF_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAF_PROD" />
</ARCS>
</NODE>
<NODE name= "reuseOAF_PROD" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF content from the PROD aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAF_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "patchRelations_PROD" />
</ARCS>
</NODE>
<NODE name= "patchRelations_PROD" type= "SetEnvParameter" >
<DESCRIPTION > should apply the relations id patching based on the provided idMapping on PROD?</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > shouldPatchRelations_PROD</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > false</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "contentPathProd" />
</ARCS>
</NODE>
<NODE name= "contentPathProd" type= "SetEnvParameter" >
<DESCRIPTION > set the PROD aggregator content path</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > prodContentPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/prod_aggregator_for_beta</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "prodAggregatorGraphPath" />
</ARCS>
</NODE>
<NODE name= "prodAggregatorGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the path containing the PROD AGGREGATOR graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > prodAggregatorGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/00_prod_graph_aggregator</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig2" />
</ARCS>
</NODE>
<NODE name= "reuseODFClaims_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF claims from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODFClaims_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseODF_hdfs_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseODF_hdfs_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF records on HDFS from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODFhdfs_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAFClaims_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseOAFClaims_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF claims from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAFClaims_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAF_hdfs_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseOAF_hdfs_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF records on HDFS from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAFhdfs_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseDB_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseDB_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached DB content from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseDB_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseDBOpenorgs_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseDBOpenorgs_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OpenOrgs content from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseDBOpenorgs_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseODF_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseODF_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached ODF content from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseODF_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "reuseOAF_BETA" />
</ARCS>
</NODE>
<NODE name= "reuseOAF_BETA" type= "SetEnvParameter" >
<DESCRIPTION > reuse cached OAF content from the BETA aggregation system</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > reuseOAF_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "patchRelations_BETA" />
</ARCS>
</NODE>
<NODE name= "patchRelations_BETA" type= "SetEnvParameter" >
<DESCRIPTION > should apply the relations id patching based on the provided idMapping on BETA?</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > shouldPatchRelations_BETA</PARAM>
<PARAM function= "validValues(['true', 'false'])" managedBy= "user" name= "parameterValue" required= "true" type= "string" > true</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "contentPathBeta" />
</ARCS>
</NODE>
<NODE name= "contentPathBeta" type= "SetEnvParameter" >
<DESCRIPTION > set the BETA aggregator content path</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > betaContentPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_aggregator</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "betaAggregatorGraphPath" />
</ARCS>
</NODE>
<NODE name= "betaAggregatorGraphPath" type= "SetEnvParameter" >
<DESCRIPTION > Set the path containing the BETA AGGREGATOR graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "parameterName" required= "true" type= "string" > betaAggregatorGraphPath</PARAM>
<PARAM managedBy= "user" name= "parameterValue" required= "true" type= "string" > /tmp/beta_experiment/graph/00_beta_graph_aggregator</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitConfig2" />
</ARCS>
</NODE>
<NODE isJoin= "true" name= "waitConfig2" >
<DESCRIPTION > wait configurations</DESCRIPTION>
<PARAMETERS />
<ARCS >
<ARC to= "betaAggregatorGraph" />
<ARC to= "prodAggregatorGraph" />
</ARCS>
</NODE>
<NODE name= "betaAggregatorGraph" type= "SubmitHadoopJob" >
<DESCRIPTION > create the BETA AGGREGATOR graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphOutputPath' : 'betaAggregatorGraphPath',
'isLookupUrl' : 'isLookUpUrl',
'reuseODFClaims' : 'reuseODFClaims_BETA',
'reuseOAFClaims' : 'reuseOAFClaims_BETA',
'reuseDB' : 'reuseDB_BETA',
'reuseDBOpenorgs' : 'reuseDBOpenorgs_BETA',
'reuseODF' : 'reuseODF_BETA',
'reuseODF_hdfs' : 'reuseODFhdfs_BETA',
'reuseOAF' : 'reuseOAF_BETA',
'reuseOAF_hdfs' : 'reuseOAFhdfs_BETA',
'contentPath' : 'betaContentPath',
'nsPrefixBlacklist' : 'nsPrefixBlacklist_BETA',
'shouldPatchRelations' : 'shouldPatchRelations_BETA',
'idMappingPath' : 'idMappingPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/raw_all/oozie_app',
'mongoURL' : 'mongodb://beta.services.openaire.eu',
'mongoDb' : 'mdstore',
'mdstoreManagerUrl' : 'https://beta.services.openaire.eu/mdstoremanager',
2022-01-21 13:59:46 +01:00
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus',
2022-01-03 14:33:00 +01:00
'postgresUser' : '',
'postgresPassword' : '',
2022-01-21 13:59:46 +01:00
'postgresOpenOrgsURL' : 'jdbc:postgresql://10.19.65.40:5432/oa_organizations',
2022-01-03 14:33:00 +01:00
'postgresOpenOrgsUser' : '',
'postgresOpenOrgsPassword' : '',
'shouldHashId' : 'true',
2022-01-21 13:59:46 +01:00
'importOpenorgs' : 'false',
2022-01-03 14:33:00 +01:00
'workingDir' : '/tmp/beta_experiment/working_dir/beta_aggregator'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitAggregatorGraph" />
</ARCS>
</NODE>
<NODE name= "prodAggregatorGraph" type= "SubmitHadoopJob" >
<DESCRIPTION > create the PROD AGGREGATOR graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphOutputPath' : 'prodAggregatorGraphPath',
'isLookupUrl' : 'isLookUpUrl',
'reuseODFClaims' : 'reuseODFClaims_PROD',
'reuseOAFClaims' : 'reuseOAFClaims_PROD',
'reuseDB' : 'reuseDB_PROD',
'reuseDBOpenorgs' : 'reuseDBOpenorgs_PROD',
'reuseODF' : 'reuseODF_PROD',
'reuseODF_hdfs' : 'reuseODFhdfs_PROD',
'reuseOAF' : 'reuseOAF_PROD',
'reuseOAF_hdfs' : 'reuseOAFhdfs_PROD',
'contentPath' : 'prodContentPath',
'nsPrefixBlacklist' : 'nsPrefixBlacklist_PROD',
'shouldPatchRelations' : 'shouldPatchRelations_PROD',
'idMappingPath' : 'idMappingPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/raw_all/oozie_app',
'mongoURL' : 'mongodb://services.openaire.eu',
'mongoDb' : 'mdstore',
'mdstoreManagerUrl' : 'https://services.openaire.eu/mdstoremanager',
2022-01-21 13:59:46 +01:00
'postgresURL' : 'jdbc:postgresql://postgresql.services.openaire.eu:5432/dnet_openaireplus',
2022-01-03 14:33:00 +01:00
'postgresUser' : '',
'postgresPassword' : '',
2022-01-21 13:59:46 +01:00
'postgresOpenOrgsURL' : 'jdbc:postgresql://10.19.65.39:5432/oa_organizations',
2022-01-03 14:33:00 +01:00
'postgresOpenOrgsUser' : '',
'postgresOpenOrgsPassword' : '',
'shouldHashId' : 'true',
'importOpenorgs' : 'true',
'workingDir' : '/tmp/beta_experiment/working_dir/prod_aggregator'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "waitAggregatorGraph" />
</ARCS>
</NODE>
<NODE isJoin= "true" name= "waitAggregatorGraph" >
<DESCRIPTION > wait configurations</DESCRIPTION>
<PARAMETERS />
<ARCS >
<ARC to= "mergeAggregatorGraphs" />
</ARCS>
</NODE>
<NODE name= "mergeAggregatorGraphs" type= "SubmitHadoopJob" >
<DESCRIPTION > create the AGGREGATOR graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'betaInputGraphPath' : 'betaAggregatorGraphPath',
'prodInputGraphPath' : 'prodAggregatorGraphPath',
'graphOutputPath' : 'mergedGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/merge/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/merge_graph',
'priority' : 'BETA'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "promoteActionsRaw" />
</ARCS>
</NODE>
<NODE name= "promoteActionsRaw" type= "SubmitHadoopJob" >
<DESCRIPTION > create the RAW graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'inputActionSetIds' : 'actionSetIdsRawGraph',
'inputGraphRootPath' : 'mergedGraphPath',
'outputGraphRootPath' : 'rawGraphPath',
'isLookupUrl' : 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/actionmanager/wf/main/oozie_app',
'sparkExecutorCores' : '3',
'sparkExecutorMemory' : '10G',
'activePromoteDatasetActionPayload' : 'true',
'activePromoteDatasourceActionPayload' : 'true',
'activePromoteOrganizationActionPayload' : 'true',
'activePromoteOtherResearchProductActionPayload' : 'true',
'activePromoteProjectActionPayload' : 'true',
'activePromotePublicationActionPayload' : 'true',
'activePromoteRelationActionPayload' : 'true',
'activePromoteResultActionPayload' : 'true',
'activePromoteSoftwareActionPayload' : 'true',
'mergeAndGetStrategy' : 'MERGE_FROM_AND_GET',
'workingDir' : '/tmp/beta_experiment/working_dir/promoteActionsRaw'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "graphCleaningFirst" />
</ARCS>
</NODE>
<NODE name= "graphCleaningFirst" type= "SubmitHadoopJob" >
<DESCRIPTION > clean the properties in the graph typed as Qualifier according to the vocabulary indicated in schemeid</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphInputPath' : 'rawGraphPath',
'graphOutputPath': 'cleanedFirstGraphPath',
'isLookupUrl': 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/clean/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/clean'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "patchHostedBy" />
</ARCS>
</NODE>
<NODE name= "patchHostedBy" type= "SubmitHadoopJob" >
<DESCRIPTION > updates publication's hostedby info according to the ISSNs available from DOAJ and UNIBI</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'cleanedFirstGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'resumeFrom' : 'produceHBM',
'hostedByMapPath' : '/user/dnet.beta/data/hostedByMap',
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/hostedbymap/oozie_app',
'sparkExecutorCores' : '2',
'sparkExecutorMemory' : '12G',
'workingDir' : '/tmp/beta_experiment/working_dir/hostedbymap',
'outputPath' : '/tmp/beta_experiment/working_dir/hostedbymap'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "graphResolution" />
</ARCS>
</NODE>
<NODE name= "graphResolution" type= "SubmitHadoopJob" >
<DESCRIPTION > Resolve Relation</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
2022-01-21 13:59:46 +01:00
{
2022-01-03 14:33:00 +01:00
'graphBasePath':'cleanedFirstGraphPath',
'unresolvedPath' :'unresolvedEntityPath',
'targetPath':'resolvedGraphPath'
2022-01-21 13:59:46 +01:00
}
2022-01-03 14:33:00 +01:00
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/resolution/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/relation_resolution',
'sparkExecutorCores' : '2',
'sparkExecutorMemory' : '12G'
}
</PARAM>
</PARAMETERS>
2022-01-21 13:59:46 +01:00
<ARCS >
<ARC to= "graphEntityGrouping" />
</ARCS>
</NODE>
<NODE name= "graphEntityGrouping" type= "SubmitHadoopJob" >
<DESCRIPTION > Resolve Relation</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphBasePath':'resolvedGraphPath',
'targetPath':'groupedGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/group/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/grouping',
'sparkExecutorCores' : '4',
'sparkExecutorMemory' : '7G'
}
</PARAM>
</PARAMETERS>
2022-01-03 14:33:00 +01:00
<ARCS >
<ARC to= "prepareIISActionSets" />
</ARCS>
</NODE>
<NODE name= "prepareIISActionSets" type= "PrepareActionSets" >
<DESCRIPTION > prepare IIS action sets</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "sets" required= "true" type= "string" >
[
{
'set' : 'iis-document-affiliation',
'jobProperty' : 'export_action_set_id_matched_doc_organizations',
'enablingProperty' : 'active_document_affiliation',
'enabled' : 'true'
},
{
'set' : 'iis-referenced-projects-main',
'jobProperty' : 'export_action_set_id_document_referencedProjects',
'enablingProperty' : 'active_referenceextraction_project',
'enabled' : 'true'
},
{
'set' : 'iis-referenced-datasets-main',
'jobProperty' : 'export_action_set_id_document_referencedDatasets',
'enablingProperty' : 'active_referenceextraction_dataset',
'enabled' : 'true'
},
{
'set' : 'iis-researchinitiative',
'jobProperty' : 'export_action_set_id_document_research_initiative',
'enablingProperty' : 'active_referenceextraction_researchinitiative',
'enabled' : 'true'
},
{
'set' : 'iis-document-similarities',
'jobProperty' : 'export_action_set_id_document_similarities_standard',
'enablingProperty' : 'active_documentssimilarity',
'enabled' : 'true'
},
{
'set' : 'iis-document-classes',
'jobProperty' : 'export_action_set_id_document_classes',
'enablingProperty' : 'active_documentsclassification',
'enabled' : 'true'
},
{
'set' : 'iis-document-citations',
'jobProperty' : 'export_action_set_id_document_referencedDocuments',
'enablingProperty' : 'active_citationmatching',
'enabled' : 'true'
},
{
'set' : 'iis-document-citations-relations',
'jobProperty' : 'export_action_set_id_citation_relations',
'enablingProperty' : 'active_citationmatching_relations',
'enabled' : 'true'
},
{
'set' : 'iis-referenceextraction-pdb',
'jobProperty' : 'export_action_set_id_document_pdb',
'enablingProperty' : 'active_referenceextraction_pdb',
'enabled' : 'true'
},
{
'set' : 'document_software_url',
'jobProperty' : 'export_action_set_id_document_software_url',
'enablingProperty' : 'active_referenceextraction_software_url',
'enabled' : 'true'
},
{
'set' : 'iis-entities-software',
'jobProperty' : 'export_action_set_id_entity_software',
'enablingProperty' : 'active_referenceextraction_software_url',
'enabled' : 'true'
},
{
'set' : 'iis-communities',
'jobProperty' : 'export_action_set_id_document_community',
'enablingProperty' : 'active_referenceextraction_community',
'enabled' : 'true'
},
{
'set' : 'iis-referenced-patents',
'jobProperty' : 'export_action_set_id_document_patent',
'enablingProperty' : 'active_referenceextraction_patent',
'enabled' : 'true'
},
{
'set' : 'iis-entities-patent',
'jobProperty' : 'export_action_set_id_entity_patent',
'enablingProperty' : 'active_referenceextraction_patent',
'enabled' : 'true'
},
{
'set' : 'iis-covid-19',
'jobProperty' : 'export_action_set_id_document_covid19',
'enablingProperty' : 'active_referenceextraction_covid19',
'enabled' : 'true'
}
]
</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "prepareIISParameters" />
</ARCS>
</NODE>
<NODE name= "prepareIISParameters" type= "PrepareIisMainParamsV2" >
<DESCRIPTION > prepare IIS parameters</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "islookupLocationParam" required= "true" type= "string" > import_islookup_service_location</PARAM>
<PARAM managedBy= "system" name= "objectStoreParam" required= "true" type= "string" > import_content_objectstores_csv</PARAM>
<PARAM managedBy= "system" name= "objectStoreLocationParam" required= "true" type= "string" > import_content_object_store_location</PARAM>
<PARAM managedBy= "system" name= "mdStoreStoreLocationParam" required= "true" type= "string" > import_mdstore_service_location</PARAM>
<PARAM managedBy= "system" name= "mdStoreDatasetParam" required= "true" type= "string" > import_dataset_mdstore_ids_csv</PARAM>
<PARAM managedBy= "system" name= "oozieWfAppPathParam" required= "true" type= "string" > oozie.wf.application.path</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "oozieWfAppPath" required= "true" type= "string" > /lib/iis/primary/snapshots/2021-12-09</PARAM>
2022-01-03 14:33:00 +01:00
<PARAM function= "validValues(['IIS','DM'])" managedBy= "user" name= "clusterName" required= "true" type= "string" > IIS</PARAM>
2022-01-21 13:59:46 +01:00
<PARAM managedBy= "user" name= "importHbaseDumpLocation" required= "true" type= "string" > deprecated - not used</PARAM>
2022-01-03 14:33:00 +01:00
<PARAM managedBy= "system" name= "importHbaseDumpLocationParam" required= "true" type= "string" > import_infospace_graph_location</PARAM>
<PARAM managedBy= "user" name= "objectStoreBlacklistCSV" required= "false" type= "string" />
<PARAM managedBy= "system" name= "importProjectConceptsContextCSVParam" required= "true" type= "string" > import_project_concepts_context_ids_csv</PARAM>
<PARAM managedBy= "user" name= "importProjectConceptsContextCSV" required= "true" type= "string" > aginfra,beopen,citizen-science,clarin,covid-19,dariah,dh-ch,oa-pg,egi,elixir-gr,embrc,enermaps,epos,euromarine,fet-fp7,fet-h2020,fam,galaxy,gotriple,ifremer,inspired-ris,instruct,mes,neanias-underwater,ni,rda,science-innovation-policy,risis,rural-digital-europe,sdsn-gr,sobigdata</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "IIS_main" />
</ARCS>
</NODE>
<NODE name= "IIS_main" type= "SubmitHadoopJob" >
<DESCRIPTION > IIS main</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > iisMainJobV3</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'cluster' : 'cluster',
'oozie.wf.application.path' : 'oozie.wf.application.path',
'active_document_affiliation' : 'active_document_affiliation',
'active_referenceextraction_project' : 'active_referenceextraction_project',
'active_referenceextraction_dataset' : 'active_referenceextraction_dataset',
'active_referenceextraction_researchinitiative' : 'active_referenceextraction_researchinitiative',
'active_documentsclassification' : 'active_documentsclassification',
'active_documentssimilarity' : 'active_documentssimilarity',
'active_citationmatching' : 'active_citationmatching',
'active_citationmatching_relations' : 'active_citationmatching_relations',
'active_referenceextraction_pdb' : 'active_referenceextraction_pdb',
'active_referenceextraction_software_url' : 'active_referenceextraction_software_url',
'active_referenceextraction_community' : 'active_referenceextraction_community',
'active_referenceextraction_patent' : 'active_referenceextraction_patent',
'active_referenceextraction_covid19' : 'active_referenceextraction_covid19',
'import_content_objectstores_csv' : 'import_content_objectstores_csv',
'import_content_object_store_location' : 'import_content_object_store_location',
'import_mdstore_service_location' : 'import_mdstore_service_location',
'import_islookup_service_location' : 'import_islookup_service_location',
'import_project_concepts_context_ids_csv' : 'import_project_concepts_context_ids_csv',
'import_dataset_mdstore_ids_csv' : 'import_dataset_mdstore_ids_csv',
2022-01-21 13:59:46 +01:00
'import_infospace_graph_location' : 'groupedGraphPath',
2022-01-03 14:33:00 +01:00
'export_action_set_id_matched_doc_organizations' : 'export_action_set_id_matched_doc_organizations',
'export_action_set_id_document_referencedDatasets' : 'export_action_set_id_document_referencedDatasets',
'export_action_set_id_document_referencedProjects' : 'export_action_set_id_document_referencedProjects',
'export_action_set_id_document_research_initiative' : 'export_action_set_id_document_research_initiative',
'export_action_set_id_document_similarities_standard' : 'export_action_set_id_document_similarities_standard',
'export_action_set_id_document_referencedDocuments' : 'export_action_set_id_document_referencedDocuments',
'export_action_set_id_citation_relations' : 'export_action_set_id_citation_relations',
'export_action_set_id_document_pdb' : 'export_action_set_id_document_pdb',
'export_action_set_id_document_software_url' : 'export_action_set_id_document_software_url',
'export_action_set_id_entity_software' : 'export_action_set_id_entity_software',
'export_action_set_id_document_community' : 'export_action_set_id_document_community',
'export_action_set_id_document_patent' : 'export_action_set_id_document_patent',
'export_action_set_id_entity_patent' : 'export_action_set_id_entity_patent',
'export_action_set_id_document_covid19' : 'export_action_set_id_document_covid19',
'export_action_set_id_document_classes' : 'export_action_set_id_document_classes'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'import_mdstore_service_location' : 'http://services.openaire.eu:8280/is/services/mdStore',
'import_content_object_store_location' : 'http://services.openaire.eu:8280/is/services/objectStore',
'import_islookup_service_location' : 'http://services.openaire.eu:8280/is/services/isLookUp',
'import_content_objectstores_csv': '258755af-0b48-41ee-9652-939c5bd2fca3_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,2ad5f567-386d-4812-8edb-c0922eacd107_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,f3b32c75-9077-4788-83ee-c8451215043c_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,7e780872-eff7-4a03-b5a2-13ad69c01366_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,0a0fcd82-ae5c-4a10-af47-5106e881b639_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,c16030ef-917e-462a-8414-2a8e2fac7619_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,8f3fdd6e-a2cd-49ad-ba93-73c8184190bf_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,909b5ec2-6859-4acf-a86e-22e31933392b_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,cea77c2a-feba-4c02-ac3c-d51a159ec904_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,2db71b6a-4838-4c25-9883-d4689148bb2a_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,8d7df838-62c9-4adc-8cd3-f7e84f732ff4_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,f90b105b-d12e-4061-99d3-0fdb85d10258_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,3e74e2c5-4844-40a4-b85f-b3ef920612b9_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,ddd7b3cc-01c2-4512-9550-52b2446f7dfa_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,2a09fb42-38f0-4a7a-a95d-97d81f716e77_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,83b7328f-2ab7-44eb-b213-2d770c4f074e_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,99d7d7f0-4f76-4cb5-9b22-3ee8ec28b9c9_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,4a484614-fd61-4ca5-b520-610335065fa2_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,32a76449-27eb-4b5f-b0e9-11cd9af0e035_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,9d22aa91-83c6-44d4-b614-943d855734a5_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,71ddf676-f6a4-48c9-9d1d-4ca6742e7316_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,d1f4546f-e47c-47c7-b19c-6334633b5f7e_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,03ebfd29-88a6-4757-9d05-64ff5018b91a_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,ba437ca2-8071-4835-9615-a14eb1fb9ff8_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,3cf2373a-8228-4781-b0f2-9b4343487fc9_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,7818e488-0ba4-4558-8f76-83cacf2f7358_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,fcfcbfd8-4e7a-46a8-a5ff-281c5fa11767_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,7b988985-c798-44f1-bff1-5574d2abe8af_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,24d97463-144b-4bb5-b6b6-7f1f5b014833_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,5b84e8f6-3ab9-40ad-be6f-7b60b24b9e42_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,a8c1d0ec-9f3e-42ba-8748-f9b5e62eb7a1_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,3cd6626e-8e0c-49ae-89a7-5e1286bddf8e_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,097344eb-7312-4096-ab9d-204c6fd8395e_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,42e09e1f-352b-4ba4-98e9-0826635009d9_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,792e67b2-29ed-40ac-a406-61726f5921ab_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,f78c1ada-b747-4862-a68b-45d61613199f_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,5707d04f-96b0-4e59-8607-5d095a3bd301_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,8fce096c-c0d8-410d-9f07-a93ac87cf2a6_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,2f817fb6-3b23-474a-b137-0c4739c264b6_T2JqZWN0U3RvcmVEU1Jlc291cmNlcy9PYmplY3RTdG9yZURTUmVzb3VyY2VUeXBl,3a5fb37f-8d49-4432-a294-538af185ac88_T2JqZWN0U3RvcmVEU1Jl
}
</PARAM>
<PARAM managedBy= "user" name= "simulation" required= "true" type= "boolean" > false</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "updateIISActionSets" />
</ARCS>
</NODE>
<NODE name= "updateIISActionSets" type= "UpdateActionSets" >
<DESCRIPTION > update IIS action sets</DESCRIPTION>
<PARAMETERS />
<ARCS >
<ARC to= "promoteActionsIIS" />
</ARCS>
</NODE>
<NODE name= "promoteActionsIIS" type= "SubmitHadoopJob" >
<DESCRIPTION > create the INFERRED graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'inputActionSetIds' : 'actionSetIdsIISGraph',
2022-01-21 13:59:46 +01:00
'inputGraphRootPath' : 'groupedGraphPath',
2022-01-03 14:33:00 +01:00
'outputGraphRootPath' : 'inferredGraphPath',
'isLookupUrl' : 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/actionmanager/wf/main/oozie_app',
'sparkExecutorCores' : '3',
'sparkExecutorMemory' : '10G',
'activePromoteDatasetActionPayload' : 'true',
'activePromoteDatasourceActionPayload' : 'true',
'activePromoteOrganizationActionPayload' : 'true',
'activePromoteOtherResearchProductActionPayload' : 'true',
'activePromoteProjectActionPayload' : 'true',
'activePromotePublicationActionPayload' : 'true',
'activePromoteRelationActionPayload' : 'true',
'activePromoteResultActionPayload' : 'true',
'activePromoteSoftwareActionPayload' : 'true',
'mergeAndGetStrategy' : 'MERGE_FROM_AND_GET',
'workingDir' : '/tmp/beta_experiment/working_dir/promoteActionsIIS'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "duplicateScan" />
</ARCS>
</NODE>
<NODE name= "duplicateScan" type= "SubmitHadoopJob" >
<DESCRIPTION > search for duplicates in the raw graph</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'actionSetId' : 'dedupConfig',
'graphBasePath' : 'inferredGraphPath',
'dedupGraphPath': 'dedupGraphPath',
'isLookUpUrl' : 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/dedup/scan/oozie_app',
'actionSetIdOpenorgs' : 'dedup-similarity-organization-simple',
'workingPath' : '/tmp/beta_experiment/working_dir/dedup',
'whiteListPath' : '/data/dedup/whitelist_beta',
'sparkExecutorCores' : '4',
'sparkExecutorMemory' : '7G',
'sparkDriverMemory' : '6G'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "dedupConsistency" />
</ARCS>
</NODE>
<NODE name= "dedupConsistency" type= "SubmitHadoopJob" >
<DESCRIPTION > mark duplicates as deleted and redistribute the relationships</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphBasePath' : 'dedupGraphPath',
'graphOutputPath': 'consistentGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/dedup/consistency/oozie_app',
'workingPath' : '/tmp/beta_experiment/working_dir/dedup'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "orcidPropagation" />
</ARCS>
</NODE>
<NODE name= "orcidPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > propagates ORCID among results linked by allowedsemrels semantic relationships</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'consistentGraphPath',
'outputPath': 'orcidGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/orcidtoresultfromsemrel/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/orcid',
'allowedsemrels' : 'IsSupplementedBy;IsSupplementTo;isSupplementedBy;isSupplementTo',
'saveGraph' : 'true',
'sparkExecutorCores' : '3',
'sparkExecutorMemory' : '10G'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "bulkTagging" />
</ARCS>
</NODE>
<NODE name= "bulkTagging" type= "SubmitHadoopJob" >
<DESCRIPTION > mark results respecting some rules as belonging to communities</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'orcidGraphPath',
'outputPath': 'bulkTaggingGraphPath',
'isLookUpUrl' : 'isLookUpUrl',
'pathMap' : 'bulkTaggingPathMap'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/bulktag/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/bulktag'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "affiliationPropagation" />
</ARCS>
</NODE>
<NODE name= "affiliationPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > creates relashionships between results and organizations when the organizations are associated to institutional repositories</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'bulkTaggingGraphPath',
'outputPath': 'affiliationGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/affiliation/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/affiliation',
'saveGraph' : 'true',
'blacklist' : 'empty'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "affiliationSemRelPropagation" />
</ARCS>
</NODE>
<NODE name= "affiliationSemRelPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > creates relashionships between results and organizations when the organizations are in parent/child relationships. The childs to the parents</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'affiliationGraphPath',
2022-01-21 13:59:46 +01:00
'outputPath': 'affiliationSemRelGraphPath',
'iterations':'iterations'
2022-01-03 14:33:00 +01:00
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/affiliation_semrel/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/affiliationsemrel'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "communityOrganizationPropagation" />
</ARCS>
</NODE>
<NODE name= "communityOrganizationPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > marks as belonging to communities the result collected from datasources related to the organizations specified in the organizationCommunityMap</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'affiliationSemRelGraphPath',
'outputPath': 'communityOrganizationGraphPath',
'organizationtoresultcommunitymap': 'propagationOrganizationCommunityMap'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/community_organization/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/community_organization',
'saveGraph' : 'true'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "resultProjectPropagation" />
</ARCS>
</NODE>
<NODE name= "resultProjectPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > created relation between projects and results linked to other results trough allowedsemrel semantic relations linked to projects</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'communityOrganizationGraphPath',
'outputPath': 'fundingGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/funding/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/funding',
'allowedsemrels' : 'IsSupplementedBy;IsSupplementTo',
'saveGraph' : 'true'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "communitySemrelPropagation" />
</ARCS>
</NODE>
<NODE name= "communitySemrelPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > tag as belonging to communitites result in in allowedsemrels relation with other result already linked to communities </DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'fundingGraphPath',
'outputPath': 'communitySemRelGraphPath',
'isLookUpUrl' : 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/community_semrel/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/community_semrel',
'allowedsemrels' : 'IsSupplementedBy;IsSupplementTo',
'saveGraph' : 'true'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "countryPropagation" />
</ARCS>
</NODE>
<NODE name= "countryPropagation" type= "SubmitHadoopJob" >
<DESCRIPTION > associated to results colleced from allowedtypes and those in the whithelist the country of the organization(s) handling the datasource it is collected from </DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'communitySemRelGraphPath',
'outputPath': 'countryGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/country/oozie_app',
'sparkExecutorCores' : '3',
'sparkExecutorMemory' : '10G',
'workingDir' : '/tmp/beta_experiment/working_dir/country',
'allowedtypes' : 'pubsrepository::institutional',
'whitelist':'10|opendoar____::16e6a3326dd7d868cbc926602a61e4d0;10|eurocrisdris::fe4903425d9040f680d8610d9079ea14;10|openaire____::5b76240cc27a58c6f7ceef7d8c36660e;10|openaire____::172bbccecf8fca44ab6a6653e84cb92a;10|openaire____::149c6590f8a06b46314eed77bfca693f;10|eurocrisdris::a6026877c1a174d60f81fd71f62df1c1;10|openaire____::4692342f0992d91f9e705c26959f09e0;10|openaire____::8d529dbb05ec0284662b391789e8ae2a;10|openaire____::345c9d171ef3c5d706d08041d506428c;10|opendoar____::1c1d4df596d01da60385f0bb17a4a9e0;10|opendoar____::7a614fd06c325499f1680b9896beedeb;10|opendoar____::1ee3dfcd8a0645a25a35977997223d22;10|opendoar____::d296c101daa88a51f6ca8cfc1ac79b50;10|opendoar____::798ed7d4ee7138d49b8828958048130a;10|openaire____::c9d2209ecc4d45ba7b4ca7597acb88a2;10|eurocrisdris::c49e0fe4b9ba7b7fab717d1f0f0a674d;10|eurocrisdris::9ae43d14471c4b33661fedda6f06b539;10|eurocrisdris::432ca599953ff50cd4eeffe22faf3e48',
'saveGraph' : 'true'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "graphCleaning" />
</ARCS>
</NODE>
<NODE name= "graphCleaning" type= "SubmitHadoopJob" >
<DESCRIPTION > clean the properties in the graph typed as Qualifier according to the vocabulary indicated in schemeid</DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'graphInputPath' : 'countryGraphPath',
'graphOutputPath': 'cleanedGraphPath',
'isLookupUrl': 'isLookUpUrl'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/clean/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/clean'
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "blacklistRelations" />
</ARCS>
</NODE>
<NODE name= "blacklistRelations" type= "SubmitHadoopJob" >
<DESCRIPTION > removes blacklisted relations </DESCRIPTION>
<PARAMETERS >
<PARAM managedBy= "system" name= "hadoopJob" required= "true" type= "string" > executeOozieJob</PARAM>
<PARAM managedBy= "system" name= "cluster" required= "true" type= "string" > IIS</PARAM>
<PARAM managedBy= "system" name= "envParams" required= "true" type= "string" >
{
'sourcePath' : 'cleanedGraphPath',
'outputPath': 'blacklistedGraphPath'
}
</PARAM>
<PARAM managedBy= "system" name= "params" required= "true" type= "string" >
{
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/enrichment/blacklist/oozie_app',
'workingDir' : '/tmp/beta_experiment/working_dir/blacklist',
2022-01-21 13:59:46 +01:00
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus',
2022-01-03 14:33:00 +01:00
'postgresUser' : '',
'postgresPassword' : ''
}
</PARAM>
<PARAM managedBy= "system" name= "oozieReportActionsCsv" required= "true" type= "string" > build-report</PARAM>
</PARAMETERS>
<ARCS >
<ARC to= "success" />
</ARCS>
</NODE>
</CONFIGURATION>
<STATUS >
2022-01-21 13:59:46 +01:00
<LAST_EXECUTION_ID > wf_20220111_200505_785</LAST_EXECUTION_ID>
<LAST_EXECUTION_DATE > 2022-01-11T20:08:53+00:00</LAST_EXECUTION_DATE>
<LAST_EXECUTION_STATUS > </LAST_EXECUTION_STATUS>
<LAST_EXECUTION_ERROR > </LAST_EXECUTION_ERROR>
2022-01-03 14:33:00 +01:00
</STATUS>
</BODY>
2022-01-21 13:59:46 +01:00
</RESOURCE_PROFILE>