Similarity Check users are reporting a different Similarity Score displaying in the document viewer, vs the PDF or through the API integration.
If the file being checked is in a folder which has exclusions applied, the document viewer is generating the score as if there were no exclusions. The correct score (with the exclusions in place) is being shown in the PDF and through the API.
As a workaround, subscribers can re-apply the exclusions to the document viewer.
Turnitin are aware of the issue and are working to solve the problem.