An Accessibility Evaluation System For Web Image Alternative Text

With the development of the Internet, for better conveying information and improving user experience, more and more images appeared in the web. For people with disabilities, especially the visually impaired, when accessing the web using auxiliary tools, the information source of the web image will be the web image alternative text, thus, it is important to provide the web image alternative text which meet accessibility requirements. However, the traditional web accessibility evaluation systems do not have the evaluation of web image alternative text.In this paper, base on the fact that the web images are related to the web pages, we design and implement an accessibility evaluation system for web image alternative text, using the content of web page, to do the accessibility evaluation.We design two algorithms for this system, the algorithm of the information extraction of web image surrounding text and the algorithm of the accessibility evaluation for web image alternative text. In the algorithm of the information extraction of web image surrounding text, we use the space features of web images to select important images, use the web page segmentation and the visual feature of web image surrounding text to do the extraction. In the algorithm of the accessibility evaluation of web image alternative text, we design a multi-view topic model to calculate the semantic similarity between the alternative text and web page title, the semantic similarity between the alternative text and web content and the semantic similarity between the alternative text and web image surrounding text, using the semantic similarity as features, we do the classifition to complete the evaluation.Using the accessibility evaluation system for web image alternative text of this paper, we solve the problem that the accessibility of web image alternative text cannot be evaluated in traditional web accessibility evaluation systems.