1. Administration and Configuration Guide
  2. Entity Resolution
  3. Executing the ER Jobs
  4. Perform Matching
  5. Additional Configuration for Matching with Oracle Text (OT)

4.4.2.2 Additional Configuration for Matching with Oracle Text (OT)

Note:

This section is applicable when MATCHING_MECHANISM is set to OT.

The source data is divided into buckets (N_BUCKET_ID) for performing candidate selection. Candidate selection is matched using a DBMS job on each bucket, and multiple buckets can be processed in parallel.

The following parameters are configured with respective pipeline id in the can_sel_ot_config table of the studio schema. For example, CSA_8128.
  • CAN_SEL_MAX_JOBS: Maximum number of buckets that can run anytime during candidate selection. By default, the value is 35.
  • QUERY_LOG_LEVEL: The logging level for Oracle text SQL queries for each source data. The acceptable values are:
    • INFO: Info level shows only failed matching queries in the CAN_SEL_OT_QUERY_LOG table. By default, it is set to INFO.
    • DEBUG: Debug shows all the source data SQL queries.
  • CAN_SEL_BATCH_SIZE: The maximum bucket size in the source data. By default, it is 2000.

    Note:

    It is applicable only for graph pipelines. For Entity Resolution, see CAN_SEL_BUCKET_SIZE parameter in the Additional Configurations section.
  • BUCKET_MAX_EXEC_TIME: The maximum time in seconds for candidate selection is executed on each bucket. By default, it is 7200.

    Note:

    • For processing a larger volume of data, increase the execution time.
    • If any buckets get timed out, the process gets terminated automatically, and the user needs to re-run the matching job.
  • PARALLEL_LEVEL: Database parallel hint used to query data, index, materialized view creation, and materialized view refresh. By default, it is 8.
  • APPLY_TRANSLITERATION: This flag represents the transliteration for candidate selection. By default, it is set to N.

Cleanup Steps for Job Termination

Execution of manual cleanup is required in case of any fatal user’s error. After contacting My Oracle Support, you can perform cleanup steps. For more information about cleanup steps, see the Cleanup Steps When the Bulk Similarity Job Terminated Manually section.

For more information about parameters, see the Parameters for Entity Resolution Job execution section.