Skip to main content

Managing Synonyms

When we talk about Managing Synonyms, we’re essentially discussing the ability to provide feedback to Synonyms and Re-run the model. Existing Synonyms can be viewed and accessed to provide feedback by clicking on the Synonym count next to the Semantic Object or Concept for which the Synonyms have been created.

An example can be seen in the image below:

You can give feedback to Synonyms by upvoting or downvoting. A Synonym’s feedback works much in the same way as the Concept Mapping feedback, providing the ability to capture multiple user feedback and generate score on that basis.

Users can provide feedback on all relevant Synonyms, which can include manually created as well as predicted Synonyms. Therefore, this feedback feature also provides multi-user feedback as in the case of ad-hoc mappings.

Feedback history can be viewed - as in the case of ad-hoc mappings - by clicking on the counts next to the upvotes and downvotes.

Just as in the case of ad-hoc column mappings for Concepts, only distinct latest votes are counted, and the scores are calculated based on distinct latest upvotes over total votes.

There are two options available once you’ve upvoted or downvoted the relevant mappings:

  1. Save & Run: Saves all mappings and re-runs the Synonym algorithm, which will recalculate and generate new predictions, suppress / remove unnecessary predictions based on the feedback provided till that point of time. This happens across the Tenant and not just for that Semantic Object.

  2. Save & Close: Saves all mapping feedback and redirects to the parent screen.

Synonym predictions are only generated after running the Synonym model at least once.

Useful Note: The impact of running this will not just be to generate Synonym predictions, but also to generate column to Concept predictions (ad-hoc mappings) for these concepts.

Why is that? - That is because if A is equal to B and if B is tagged to certain columns, it follows that A must also be tagged to these columns. That is, of course, a simplistic statement, but considering multiple aspects, mapping scores etc., additional predictions can be found leading to richer data discovery.