2.7.3. Operation
2.7.3.2. Start the first clustering job
Prev  Next

2.7.3.2. Start the first clustering job

Purpose

You start a new clustering pipeline for a selected catalog and return to the clustering overview to manage the running job.

Requirements

  • The logged-in account requires the "clustering-view" and "clustering-manage" permissions. If "Start new cluster job" does not appear in the clustering overview, these permissions may be missing.

  • Check to see if the intended source catalog is available in the “Catalog” drop-down menu.

  • Resolve any warnings regarding backend readiness before submitting jobs. Make sure that all required backend services are available before starting the job.

Procedure

  1. Open "Clustering" and select "Start new cluster job" [1]. This management action is available only to administrators with the clustering-manage permission.

    The overview is the starting point for a new job and later displays its current execution status.

    The overview is the starting point for a new job and later displays its current execution status.

  2. For a standard job, enter a unique name under “Job name” [1] and select the source catalog under “Catalog” [2] based on the name displayed there. The form selects its default preset under “Preset” [3] and sets the remaining options to default values.

  3. Set "Granularity" [4], select "Max depth" [5], and decide whether to enable "Generate cluster names" [6]. Review the entire query, and then select “Start cluster job” [7].

    The numbered controls together form the request to the clustering backend. Review them as a whole, because the resulting hierarchy depends on the interplay of all settings.

    The numbered controls together form the request to the clustering backend. Review them as a whole, because the resulting hierarchy depends on the interplay of all settings.

  4. After the request has been accepted, the application returns to the overview screen. The new job appears first under “Active Jobs” [1]. Track its execution using the displayed values for catalog, preset, granularity, depth, naming, phase, and progress.

    The new active job displays the submitted settings while the backend processes the pipeline.

    The new active job displays the submitted settings while the backend processes the pipeline.

Options and Decisions in Operations

First Clustering Job – Form Controls and Operational Decisions

Control element or State

Operating Instructions

"Back"

Return to the overview without submitting the form. Use this option if you want to review the available jobs before deciding on another one.

"Job Name"

Required field. Enter a unique, descriptive name that identifies the run in the job list and distinguishes it from other jobs on the same catalog. A name consisting solely of spaces is invalid.

"Catalog"

Required field. Select the catalog whose parts are to be clustered. The display name of the catalog is shown; the application sends the corresponding catalog entry to the backend.

"Preset"

Required field. Select the parameter preset provided by the backend. Once the list is loaded, the default preset is automatically selected. Use the preset that is approved for the intended clustering method; check the tooltip if a description is available.

"Granularity"

Range 0.00 to 1.00 in increments of 0.01; default 0.50. Lower values produce coarser, wider clusters; higher values produce finer clusters. Select a value that is appropriate for the expected effort involved in checking and manual correction.

Explanation of “Granularity”

Briefly explains what the granularity scale means. Use it to check the effect of a numeric value before submitting it.

"Max depth"

Limits the number of hierarchy levels generated. The default is “unlimited levels”; the backend will then use as many levels as necessary. Fixed values from 1 to 5 limit the depth if a flatter, more easily navigable result is desired.

"Generate cluster names"

Disabled by default. Enable the option so that the pipeline also handles the naming of the clusters; leave it disabled if no names are needed or if they will be generated later in a separate process.

"Start cluster job"

Sends the complete request. The button is disabled until the job name, catalog, and preset are specified, and displays a "Waiting" status while the request is being sent.

"Cancel"

Return to the overview without starting a job. Use this to discard a configuration that has not been submitted.

Map of the Active Job

The most recently accepted job appears first under “Active Jobs.” Its card displays the catalog, selected options, current phase, and progress. A completed job later appears as a saved cluster in the overview.

Result

The backend accepts the configuration, and the overview displays an active clustering job. While the pipeline is running, the job card serves as the primary status indicator. Once complete, the result is available as saved clustering for use in subsequent workflows. .

Tips and Troubleshooting

  • If "Start cluster job" is not available, enter the missing required information. The request cannot be submitted without a selected catalog, with an empty job name, or without an available preset.

  • If the preset list is not available, the form will indicate that clustering requires a preset. Restore the availability of the presets before trying again.

  • If the backend rejects a request, the form retains the current settings and displays the error message returned. Correct the reported issue and resubmit.

  • Use "Cancel" or "Back" to discard a configuration that has not been submitted; neither action creates a job.

Prev Up Next
2.7.3.1. Check access, permissions, and availability of the backend Home 2.7.3.3. Monitor and manage clustering jobs