Steps to Implement your own Bulk Reassessment Batch Process

Batch processing must be written using the chunked batch processing architecture (see the Curam Batch Performance Mechanisms document).

Write these batch programs:

a "Chunker" batch job which identifies¹the list of cases and passes them, along with appropriate control parameters to the BatchChunker to divide them into smaller lots. This process of identifying the cases can be controlled by whatever parameters are required, such as Product ID, case status, etc.
a "Stream" job which takes one of these chunks of work and performs full reassessment of each case contained in it.

You must decide on appropriate metrics to capture during the stream processing, such as the number of cases processed, and/or the number of cases reassessed which did/did not result in a changed determination result. Your chunker and streamer must share data structures so that the stream processing can capture metrics and the chunker can accumulate them into its report.

The "Stream" job must perform a full reassessment of each CER-based case as follows²:

Figure 1. Code example to reassess a CER-based case

// class member
            @Inject
            protected DeterminationCalculatorFactory
              determinationCalculatorFactory;
            
            public void yourBatchMethod(...yourparameters...)
              throws AppException, InformationalException {
            
            // process an identified case
              final long caseID = ...;
            
              final DeterminationCalculator determinationCalculator =
                determinationCalculatorFactory
                  .newInstanceForCaseID(caseID);
            
              final DetermineEligibilityKey determineEligibiltyKey =
                new DetermineEligibilityKey();
            
              determineEligibiltyKey.caseID = caseID;
            
             /*
              * reassess the case and determine whether the decision has
              * changed
              */
              final boolean decisionChanged =
               determinationCalculator
                .hasDecisionChanged(
                  determineEligibiltyKey,
                  CASEASSESSMENTDETERMINATIONREASONEntry.SOMEREASON);
            
            }

¹ You should ensure that each case is identified at most once.

For simple database queries this is unlikely to be challenging; however for non-trivial queries you might consider using SQL's DISTINCT keyword, e.g. SELECT DISTINCT(caseID) FROM....

² You must also include:

processing to capture the metrics for your batch program; and
appropriate error handling to communicate with the chunked batch processing architecture's support for skipping cases.

This extra code is beyond the scope of this guide.