Concadia

Concadia 是我们的论文 Concadia: Tackling image accessibility with descriptive texts and context 中介绍的一个数据集，它包含 Wikipedia 图像及其各自的标题、alt 描述和图像所在的更广泛的上下文。我们使用这个语料库来明确区分描述和标题，并显示两种文本形式之间的异同。我们进一步认为，字幕和更广泛的上下文是一种重要的资源，可以为生成的描述提供信息，这些描述在 Web 上非常稀疏，但对于使图像可访问绝对至关重要。

Concadia is a dataset introduced in our paper *Concadia: Tackling Image Accessibility with Descriptive Texts and Context*, which comprises Wikipedia images alongside their corresponding captions, alt descriptions, and the broader context in which the images appear. We employ this corpus to explicitly differentiate between descriptive texts and captions, as well as to illustrate the similarities and differences between these two types of textual formats. We further argue that captions and broader context represent a critical resource for informing the generation of descriptive texts—texts that are extremely sparse on the Web yet absolutely essential for ensuring image accessibility.