Table of Contents

Search

  1. Preface
  2. Analyst Service
  3. Catalog Service
  4. Content Management Service
  5. Data Integration Service
  6. Data Integration Service Architecture
  7. Data Integration Service Management
  8. Data Integration Service Grid
  9. Data Integration Service Applications
  10. Interactive Data Preparation Service
  11. Enterprise Data Preparation Service
  12. Informatica Cluster Service
  13. Mass Ingestion Service
  14. Metadata Access Service
  15. Metadata Manager Service
  16. Model Repository Service
  17. PowerCenter Integration Service
  18. PowerCenter Integration Service Architecture
  19. High Availability for the PowerCenter Integration Service
  20. PowerCenter Repository Service
  21. PowerCenter Repository Management
  22. PowerExchange Listener Service
  23. PowerExchange Logger Service
  24. SAP BW Service
  25. Search Service
  26. System Services
  27. Test Data Manager Service
  28. Test Data Warehouse Service
  29. Web Services Hub
  30. Application Service Upgrade
  31. Appendix A: Application Service Databases
  32. Appendix B: Connecting to Databases from Windows
  33. Appendix C: Connecting to Databases
  34. Appendix D: Updating the DynamicSections Parameter of a DB2 Database

Probabilistic Models and Classifier Models

Probabilistic Models and Classifier Models

The Model Repository Service reads probabilistic model and classifier model file data from the machine that hosts the master Content Management Service in the domain. When you compile a probabilistic model or classifier model in the Developer tool, you update the model files on the master Content Management Service machine.
If a node in the domain runs a Content Management Service, the node stores local copies of the probabilistic model and classifier model files. You specify the local path to the probabilistic and classifier model files in the
NLP Options
property on the Content Management Service. The master Content Management Service synchronizes the probabilistic model and classifier model files on the domain nodes with the master Content Management Service files every 10 minutes.
To synchronize a Content Management Service machine with the current files from the master Content Management Service machine, run the following command:
infacmd cms ResyncData
The command updates the machine that hosts the new service with the probabilistic model or classifier model files from the master Content Management Service machine. When you add a Content Management Service to a domain that includes a master Content Management Service, run the ResyncData command.
You specify a single type of model file when you run the command. To synchronize probabilistic model files and classifier model files, run the command once for each type of model file.

Synchronization Operations

The master Content Management Service stores a list of the Content Management Services in the domain. When the master Content Management Service synchronizes with the domain services, the master Content Management Service copies the current model files sequentially to each domain node. If a node is unavailable, the master Content Management Service moves the node to the end of the list and synchronizes with the next node on the list. After the synchronization operation copies the files to all available Content Management Service machines, the operation ends.
To verify that a synchronization operation succeeded on a node, browse the directory structure on the node and find the probabilistic or classifier model files. Compare the files with the files on the master Content Management Service machine.
Informatica uses the following directory paths as the default locations for the files:

    [Informatica_install_directory]/tomcat/bin/ner

    [Informatica_install_directory]/tomcat/bin/classifier

The file names have the following extensions:

    Probabilistic model files:
    .ner

    Classifier model files:
    .classifier

The time required to synchronize the model files depends on the number of files on the master Content Management Service machine. The ResyncData command copies model files in batches of 15 files at a time.

0 COMMENTS

We’d like to hear from you!