ARGIRA v1.4.4: Three-Layer Uncertainty Communication Prototype for Automatic Alt-Text — Accessibility & Localization Refinement

ARGIRA v1.4.4 is an accessibility and localization refinement release of the ARGIRA prototype for automatic alt-text generation and three-layer uncertainty communication, focused on improving user-facing clarity, accessible interface text, bilingual consistency, and communication of the prototype's actual capabilities and limitations.

It is a new version of ARGIRA v1.4.3 (DOI: 10.5281/zenodo.21916695).

Accessibility and interface refinement

This release provides a further refinement of the user-facing interface, with particular attention to accessible text and the information communicated to users through visual interface elements, ARIA labels, dynamic status messages, result navigation, copy and download actions, and accessibility-oriented interactions.

The English prototype has been reviewed and refined to provide consistent English-language communication across the main user interface, result information, accessibility labels, dynamic messages, metadata panels, and explanatory content.

The Spanish-localized prototype remains aligned with the English artifact, with both variants continuing to operate as independently executable HTML prototypes rather than using an in-application language selector.

Uncertainty communication and scope

The release preserves the existing three-layer uncertainty communication approach and the distinction between the capabilities of this specific demonstration and the broader ARGIRA project.

The demonstration continues to use a statistical pixel-based heuristic and predefined templates. It does not incorporate a computer-vision model for semantic image recognition. It can estimate the likely type of an image and expose the statistical signals and uncertainty associated with that estimate, but it does not identify objects, people, scenes, or other semantic content within the image.

The generated text remains an automatically constructed type-level description rather than a semantic description of the image.

Accessibility and user interaction

The release preserves the established accessibility-only navigation behavior and the essential result path: image type, confidence, generated alt-text, copy functionality, and previous/next result navigation.

Accessible labels and dynamic status messages have been refined so that important user actions and result-state changes are communicated consistently to users, including users relying on screen readers.

The release does not introduce an in-application language selector. English and Spanish remain separate, independently executable prototype artifacts.

Accessibility verification scope

Manual developer verification with Android TalkBack carried forward from previous releases remains part of the accessibility evidence for the prototype.

Apple VoiceOver was not evaluated because an available Apple/VoiceOver testing environment was not available. This remains an untested scope, not a failed validation.

No external user study or participant evaluation was conducted for this release.

Experimental scope and continuity

ARGIRA v1.4.4 does not introduce a new image-classification algorithm, calibration model, or experimental condition.

The underlying heuristic classification, calibration logic, thresholds, embedded data, identifiers, and experimental session mechanism are preserved.

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

The purpose of this release is to support continued research, reproducibility, accessibility-oriented inspection, bilingual use, clearer communication of uncertainty and limitations, and future evaluation with additional assistive-technology environments and external users.

Package contents

prototype/argira-demo-v4.html — English prototype artifact.

prototype/argira-demo-es-v4.html — Spanish-localized prototype artifact.

README.md — release overview and usage information.

CHANGELOG.md — version history.

CITATION.cff — machine-readable citation metadata.

LICENSE.txt — CC BY-NC-SA 4.0 license text.

The release provides two independently executable prototype artifacts, in English and Spanish, with structural and functional parity between the variants.