OCDM Triple Store Rebuilding

Submitter: detwiler
[NOTE: While the approach listed here is still possible, this is not how the triple stores are presently built. See the script in /opt/fma/update to view the current approach] While this is, at present, still a bit fluid, I wanted to document the current approach to rebuilding the OCDM triple store (behind the SPARQL endpoint).
  1. navigate to $FUSEKI_HOME (/usr/local/service/fuseki-tdb/)
  2. cd to ocdm_update
  3. execute the script ocdm_fuseki_update as root (simply enter './ocdm_fuseki_update.sh' while logged in as root)
The update script presently does the following things:
  1. stops the Fuseki server (this is the SPARQL endpoint server)
  2. delete the current OCDM database
  3. delete the current Lucene index over the OCDM database
  4. load a new OCDM triple store from the ontologies in the 'ocdm' subdirectory *
  5. build a new Lucene index
  6. restart the Fuseki server
The results of running this script are logged in the 'update_logs' subdirectory. * The script builds graphs in the triple store based on an expected file naming scheme. For a file whose name does not start with "pun_", a graph name is created by appending the file name to the domain prefix "http://purl.org/sig/ont/ocdm/". So, for example, if the script encounters an ontology file named "cho.owl", it will put it in a graph named "http://purl.org/sig/ont/ocdm/cho.owl". If, however, the file starts with "pun_", the same strategy is employed after removing the "pun_" prefix. For example, "pun_cho.owl" will be placed in the graph "http://purl.org/sig/ont/ocdm/cho.owl". The intent is to place both the base ontology and its associate pun file into the same named graph.