Product DocsMenu

Runs Versus Jobs

The Text Analytics module can perform operations either as a run or as a job. The following table summarizes the similarities and the differences between the two types of operations.

Aspect Run Job
Brief definition Sequence of pipeline stages iteratively executed on a set of documents. Executes a piece of code to perform any task and does not operate on a set of documents.
Executed By the Coveo Job Scheduling (CJS) service
When to use
  • Operations must be performed on each document of a set, one at a time.

  • To initially process all indexed documents and process new, deleted, or changed indexed documents at regular intervals.

  • One shot operation such as maintenance tasks like creating a backup, checking something, granting permissions, creating tag fields in the Coveo unified index, etc.

  • To whitelist, blacklist, or normalize known values in all documents using tagging queries.

Performance
  • Dependent on document set size and the number and resources required by each pipeline

  • Changes in the pipeline normally require to reprocess all documents in the pipeline, one by one.

  • Dependent on the code of the plugin

  • Changes to whiltelists, blacklists, and normalizers can be quickly applied to the index since each modification is applied on all specified documents simultaneously using CES tagging queries.

Example: After completing a text analytics run that processed all documents in your Coveo unified index, you need to normalize or blacklist extracted items. To do so:

  1. Add the necessary terms to blacklisting and normalization files.

  2. Add the blacklisting and normalization plugins using these files to your run so upcoming scheduled executions will blacklist and normalize terms for new index documents.

  3. Use a job to efficiently perform normalization and blacklisting for the index documents that have already been processed.

People who viewed this topic also viewed