Closed gdbassett closed 11 years ago
Particularly because the CPT will be anchored to the origin, the CPT needs to be updated with the dbIds to anchor to the DB.
This will be tricky.
CPTs in "AN"s will need to be anchored to origin IDs of other ANs and AEs in the event.
ANs anchored to dbIDs will force nodes in the DB to be updated. <-- big question about how to update this.
CPTs in RNs/CNs will need to be anchored to the DB.
Big question is, when adding/deleting an edge, how do you update the CPT? It'll add a column and double the rows. (or delete the same).
May want to add a "cptreviewed" property to nodes. When the CPT is automatically updated, it can be marked 'false'. Could alternately, replace the CPT and just create an 'old-cpt' property.
Suggestion: Treat "actor" and "attribute" parents as required (an 'and') and treat "event" and "condition" parents as an 'or' (i.e. only 1 needs to exist).
CPTs will need to be validated and updated AFTER all edges are added to ensure parents are in place before we try and validate the CPT.
(Could potentially search the event to see if edges WILL BE added that will satisfy parental relationships.)
updateCPTs, fixCPTs, and validateCPT written. validateCPT needs to be tested.
Validate the CPT against the standard set forth here: http://infosecanalytics.blogspot.com/2013/03/conditional-probability-tables-in-json.html