Dataset Linking Report Workflow

The Dataset Linking Report Workflow (DATASET_LINKING_REPORT) asks the Dataset Linking service to produce a report on how similar datasets are, and how they could be linked, based on their keywords, descriptions and headlines.

Input

All tuning parameters are optional. If a parameter isn't set, the Dataset Linking service uses its own default.

Parameter	Type	Mandatory	Description
id	string (uuid)	yes	Identifier of the dataset that the report is requested for.
kw	number	no	Weight of keyword similarity in the overall similarity score.
desc	number	no	Weight of description similarity in the overall similarity score.
head	number	no	Weight of headline similarity in the overall similarity score.
th	number (0–100)	no	Similarity threshold, as a percentage, that a pair of datasets must reach to be reported as related.
keyword_method	jaccard or sbert	no	Method used to compare keywords: set overlap (Jaccard) or semantic embeddings (SBERT). Defaults to jaccard.
include_chunks	boolean	no	Whether to include the matching parts of descriptions as evidence in the report.
include_graphs	boolean	no	Whether to generate graphs while the report is built.
graphs_only_above_threshold	boolean	no	Whether to generate graphs only for pairs that reach the threshold.
workflow_process_step_information	object	yes	Tracking information for the platform process step that this run belongs to. Supplied automatically by the platform. See Platform Integration.

Tasks

flowchart LR
    A[Start Report Job] --> B{Is the report complete?}
    B -- No, check again later --> B
    B -- Yes --> C((End))

Start Report Job

This task submits a report job to the Dataset Linking service with the provided parameters, and returns the identifier of the job.

Wait for Completion

This task checks the report job's status at regular intervals until the job completes, releasing its worker slot between checks. If the job fails, the workflow stops with a failure.

Completion

When the report is complete, the workflow notifies the platform, which can then retrieve the report from the Dataset Linking service. The platform then starts the next step of the dataset's lifecycle, the Dataset Packaging workflow.