forked from D-Net/dnet-hadoop
1030 lines
65 KiB
XML
1030 lines
65 KiB
XML
<RESOURCE_PROFILE>
|
|
<HEADER>
|
|
<RESOURCE_IDENTIFIER value="b05c97e6-69b5-497d-87fd-2137d3ff2c2e_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
<RESOURCE_URI value=""/>
|
|
<DATE_OF_CREATION value="2021-08-03T13:43:44+00:00"/>
|
|
</HEADER>
|
|
<BODY>
|
|
<WORKFLOW_NAME>Graph Construction [BETA]</WORKFLOW_NAME>
|
|
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE>
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
<CONFIGURATION start="manual">
|
|
<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________,fct_________</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________</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>
|
|
<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_provision/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_provision/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_provision/graph/03_graph_cleaned</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/04_graph_dedup</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/05_graph_inferred</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/06_graph_consistent</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/07_graph_orcid</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/08_graph_bulktagging</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/09_graph_affiliation</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/10_graph_comunity_organization</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/11_graph_funding</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/12_graph_comunity_sem_rel</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/13_graph_country</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/14_graph_cleaned</PARAM>
|
|
</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>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_provision/graph/15_graph_blacklisted</PARAM>
|
|
</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">scholexplorer-dump,doiboost,orcidworks-no-doi,iis-entities-software,iis-entities-patent,datacite</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,h2020classification,bipfinder-scores</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">true</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_provision/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_provision/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' : '',
|
|
'mdstoreManagerUrl' : '',
|
|
'postgresURL' : '',
|
|
'postgresUser' : '',
|
|
'postgresPassword' : '',
|
|
'postgresOpenOrgsURL' : '',
|
|
'postgresOpenOrgsUser' : '',
|
|
'postgresOpenOrgsPassword' : '',
|
|
'shouldHashId' : 'true',
|
|
'importOpenorgs' : 'true',
|
|
'workingDir' : '/tmp/beta_provision/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' : '',
|
|
'mdstoreManagerUrl' : '',
|
|
'postgresURL' : '',
|
|
'postgresUser' : '',
|
|
'postgresPassword' : '',
|
|
'postgresOpenOrgsURL' : '',
|
|
'postgresOpenOrgsUser' : '',
|
|
'postgresOpenOrgsPassword' : '',
|
|
'shouldHashId' : 'true',
|
|
'importOpenorgs' : 'true',
|
|
'workingDir' : '/tmp/beta_provision/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_provision/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_provision/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_provision/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',
|
|
'oozie.wf.application.path' : '/lib/dnet/BETA/oa/graph/hostedbymap/oozie_app',
|
|
'workingDir' : '/tmp/beta_provision/working_dir/hostedbymap',
|
|
'outputPath' : '/tmp/beta_provision/working_dir/hostedbymap'
|
|
}
|
|
</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' : 'cleanedFirstGraphPath',
|
|
'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_provision/working_dir/dedup',
|
|
'sparkExecutorCores' : '3',
|
|
'sparkExecutorMemory' : '10G'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</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',
|
|
'inputGraphRootPath' : 'dedupGraphPath',
|
|
'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_provision/working_dir/promoteActionsIIS'
|
|
}
|
|
</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' : 'inferredGraphPath',
|
|
'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_provision/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_provision/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_provision/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_provision/working_dir/affiliation',
|
|
'saveGraph' : 'true',
|
|
'blacklist' : 'empty'
|
|
}
|
|
</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' : 'affiliationGraphPath',
|
|
'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_provision/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_provision/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_provision/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_provision/working_dir/country',
|
|
'allowedtypes' : 'pubsrepository::institutional',
|
|
'whitelist' : '10|openaire____::e783372970a1dc066ce99c673090ff88;10|opendoar____::16e6a3326dd7d868cbc926602a61e4d0',
|
|
'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_provision/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_provision/working_dir/blacklist',
|
|
'postgresURL' : '',
|
|
'postgresUser' : '',
|
|
'postgresPassword' : ''
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="success"/>
|
|
</ARCS>
|
|
</NODE>
|
|
</CONFIGURATION>
|
|
<STATUS>
|
|
<LAST_EXECUTION_ID>wf_20210803_134357_367</LAST_EXECUTION_ID>
|
|
<LAST_EXECUTION_DATE>2021-08-03T17:08:11+00:00</LAST_EXECUTION_DATE>
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
<LAST_EXECUTION_ERROR/>
|
|
</STATUS>
|
|
</BODY>
|
|
</RESOURCE_PROFILE> |