ARGIRA v1.4.7: Three-Layer Uncertainty Communication Prototype for Automatic Alt-Text — Accessibility & Refinement

ARGIRA v1.4.7 is a new release of the ARGIRA prototype for automatic alt-text generation and three-layer uncertainty communication. It extends the user-facing interface and refines the accessibility behavior of the prototype for screen-reader navigation.

It is a new version of ARGIRA v1.4.6 (DOI: 10.5281/zenodo.21965506).

Language selection and cross-navigation

v1.4.7 introduces a dedicated language selector in the interface. The Spanish and English artifacts remain independently executable, while each now provides a direct link to the corresponding version hosted at argira.eus.

Screen-reader navigation and result presentation

This release refines result navigation for assistive technology. Results are presented as an ordered sequence of individually navigable cards, with explicit previous/next navigation and focus management. The result region no longer automatically announces every newly inserted card as a live-region update; users can instead reach and navigate results through the intended focus sequence.

The release also includes targeted adaptations for JAWS, Android TalkBack, and Apple VoiceOver. TalkBack verification carried forward from previous releases remains part of the developer accessibility evidence. JAWS has received a code-level adaptation and partial manual developer verification. VoiceOver has been adapted in code but has not yet been verified on a real Apple/VoiceOver environment.

These changes are accessibility-oriented implementation and developer verification work; they do not constitute a formal accessibility validation or an external user study.

Interface and content refinements

v1.4.7 also includes:

• reordered interface controls and related text for more coherent navigation;
• updated links between the Spanish and English artifacts;
• removal of internal/prototype wording from the user-facing header;
• revised user-facing error and status wording;
• preservation of the existing pagination model for navigating analyzed images.

Experimental scope and continuity

v1.4.7 does not introduce a new image-classification algorithm, calibration model, or experimental threshold. The underlying deterministic heuristic classification, calibration information, runner-up computation, and three-layer uncertainty communication remain part of the existing ARGIRA prototype.

The prototype continues to use deterministic image statistics and heuristics rather than neural networks or machine-learning models.

ARGIRA remains an experimental research artifact and is not presented as a finished product or as validated accessibility technology.

Accessibility verification scope

Developer inspection and manual accessibility verification remain limited to the environments explicitly documented in the release README. TalkBack evidence is carried forward from previous releases. JAWS verification is partial. Apple VoiceOver remains untested on a real Apple device or environment. No physical-keyboard evaluation or external participant study has been conducted for this release.

Package contents

The release provides two independently executable prototype artifacts:

• index_es-1_4_7.html — Spanish-localized prototype artifact.
• index_en-1_4_7.html — English prototype artifact.

The package also includes the release documentation, version history, citation metadata, license, and SHA-256 checksums.

Both HTML artifacts are self-contained and can be opened locally without installation or network access.

Continuity

Previous ARGIRA releases remain available as historical research artifacts and are not replaced by this publication.

DOI: 10.5281/zenodo.21970403