Comparing Content Similarity Scores

PromoMats content similarity scores allow you to more efficiently review derivative content. By combining the keyword and image similarity between two documents, PromoMats calculates and displays an overall similarity score in the Content Similarity panel on the Doc Info page, which can be used to facilitate tier-based review. The similarity score between documents is based on their relative similarity to all other documents in the Vault, so we recommend having at least 100 documents in the Vault where you want similarity scores to be calculated.

Configuring Content Similarity

Before you begin comparing similarity scores, you must add document types to the Content Similarity document type group.

Viewing the Content Similarity Panel

To open the Content Similarity panel on the Doc Info page, click .

To add documents to the Content Similarity panel, click . You can add a maximum of five (5) documents to the panel. Adding a document to the panel creates a new Created From relationship between the source and target documents. Click to remove a document from the panel and delete the associated Created From relationship.

The Content Similarity panel displays a thumbnail alongside the name, number, version, and lifecycle state for each target document. Once calculated, Vault displays the similarity score below the thumbnail when the page is refreshed.

Vault recalculates the similarity score when the source document is re-rendered.

Configuring Lifecycles & Workflows

You can use similarity scores to drive tier-based review workflows by adding the AppSimilarityScore(Document.id__v) app function to document lifecycles and document workflows. For example, you can use this function in user actions and workflow start steps and anywhere else formulas are accepted. This function searches Content Similarity records linked to the source document and returns the highest overall similarity score. 

To use this function, add an argument to the end using a decimal number. Do not remove or replace the content in parentheses. For example, if you would like to specify a behavior when a document has a similarity score greater than 90%, the formula should be AppSimilarityScore(Document.id__v)> 0.9.

Limitations

The following limitations apply to content similarity scores:

• Vault calculates similarity scores using only the first 15 pages of larger documents.
• You cannot add documents to the Content Similarity panel if the document is in its steady state.
• You cannot compare binders, audio files, and video files that do not have a viewable PDF rendition.
• You cannot compare documents that do not have indexed text.
• You cannot compare similarity scores for the following document types:
  • eCTD Submission Ready Documents (ectd_submission_ready_documents__v)
  • Reference (reference__v)
  • Template (template__v)
  • Content Module (content_module__v)
  • Event Material (event_material__v)
  • InDesign Component (indesign_component__v)
  • Undefined (undefined__v)
  • Health Authority Form (health_authority_form__v)
  • Compliance Package (compliance_package__v)
  • Engage Presentation (engage_presentation__v)
  • Template Fragment (template_fragment__v)

Related Permissions

You must have the View Content permission on target documents to view document metadata, including the similarity score, in the Content Similarity panel. You must also have the Edit Relationships permission on the source document to add or remove documents from the Content Similarity panel.