specify hierarchy/algorithm to handle insertion conflicts

monthly automatic updates of the database

• a full backup of the database should be stored before everything else
• please note that for keeping the database automatically up to date:
  • libraries should be harvested for updates since the last time we harvested them
    • for literature publications: see below
  • the enricher incl. SCHOLIX links and FAIR meter should run for all of them again
    • for data publications (dataset): see below
    • FAIR scores shall be overwritten (table needs to be extended with additional field timestamp)
      • the previous FAIR scores are appended to table fuji_score_archive for this publicationID (which still needs to be created in the database)

literature publications

• if a publication with the same PID exists already in the database:
  • check if combination of center and publication is listed in the table center_has_publication
    • if yes: check if additional metadata is available (e.g., previously empty fields and relations can be augmented) and, if yes, do that
    • if not: add this combination to the table (talk with Mojeeb)
  • check if combination of research field and publication exists in research_field_has_publication
    • if not: add this combination to the table (talk with Mojeeb)
• in every case: follow the instruction below also for "data publications (datasets)"

data publications (datasets)

• if a publication with the same PID exists already in the database:
  • check if combination of center and publication is listed in the table center_has_publication
    • if yes: check if additional metadata is available (e.g., previously empty fields and relations can be augmented) and, if yes, do that
    • if not: add this combination to the table (talk with Mojeeb)
• in every case: follow the instruction below also for "data publications (datasets)"

relationships (datasets)

• if a relation (SCHOLIX-link) between a literature publication and the data publication (dataset) with the same combination of PIDs exists already in the database:
  • check if the relationshipType is the same as the update
    • if yes: check if additional metadata is available (e.g., previously empty fields and relations can be augmented) and, if yes, do that
    • if not: update the relationship according to the following hierarchy, always keeping the strongest
      • from strongest to weakest: IsSupplementedBy - isReferencedBy - IsRelatedTo - (everything else -> log new types!)