We appreciate the opportunity to comment on the String Similarity Evaluation Guidelines for the New gTLD Program: Next Round.

With regard to the potential addition of the Armenian small letter Co (ց, U+0581) and Armenian small letter Za (զ, U+0566) as a confusingly similar pair in the similarity data, we acknowledge that under certain visual contexts, particularly when underlining is applied, these characters may appear confusingly similar. However, we understand that adding this specific pair directly to the similarity data could introduce unintended transitivity effects across other scripts, which may not be desirable from a systemic and cross-script consistency perspective.

In this context, we note that the Guidelines already provide a mechanism to address such script-specific visual considerations. In particular, Section 2 of the Guidelines explicitly allows the evaluation panel to take into account additional factors beyond predefined similarity data when assessing potential visual confusion.

As an alternative to adding this Armenian character pair to the similarity dataset, we suggest explicitly recognizing this case as an example where script-specific typographic features—such as underlining—should be considered during evaluation. The Armenian script contains characters whose visual similarity may be context-dependent rather than inherent, and this should be reflected through expert panel judgment rather than rigid data expansion.

We believe this approach strikes an appropriate balance between maintaining the integrity of the similarity data and ensuring that legitimate script-specific concerns, including those relevant to Armenian, are adequately addressed in the evaluation process.

Thank you for considering this input.

The comment supports the ICANN String Similarity Evaluation Guidelines while cautioning against adding the Armenian letters ց (U+0581) and զ (U+0566) as a confusingly similar pair in the similarity data. Although these characters may appear similar in certain contexts, such as when underlined, directly including them could create unintended transitivity effects across scripts. Instead, the submission highlights that the existing Guidelines—particularly Section 2—already allow expert panels to consider context-specific visual factors. It recommends addressing this Armenian-specific case through panel judgment rather than expanding the similarity dataset, thereby preserving systemic consistency while ensuring script-specific concerns are properly evaluated.