An optimal sampling method for web accessibility quantitative metric

Mengni Zhang,Can Wang,Jiajun Bu,Zhi Yu,Yi Lu,Ruijie Zhang,Chun Chen
DOI: https://doi.org/10.1145/2745555.2746663
2015-01-01
Abstract:When evaluating the accessibility of a website, we usually resort to sampling methods to reduce the cost of evaluation. In this kind of approaches, a small subset of pages in a website are chosen for evaluating the accessibility value of the whole website. Good sampling quality means the selected subset can represent well the accessibility level of the whole website, i.e. minimizing the accessibility evaluation difference between the whole site and the sampled subset. As existing studies show the accuracy of sampling methods depends heavily on the metric, we propose in this paper a specific sampling method OPS-WAQM that is optimized for Web Accessibility Quantitative Metric (WAQM). OPS-WAQM minimizes the sampling error by choosing the optimal sample numbers in different page depth layers. A greedy algorithm is proposed to approximately solve the optimization problem in an efficient way. We use a dataset of 20 websites, 365780 web pages to validate our method. Experimental results show that our sampling method is effective for web accessibility evaluation.
What problem does this paper attempt to address?