Considering People with Disabilities as Überusers for Eliciting Generalisable Coping Strategies on the Web

When users struggle on the Web they employ extreme adaptations to tackle problematic situations, namely coping strategies. If we are able to automatically detect such situations we can provide the means to bypass or pre-empt them. However, isolating these coping strategies is a challenging task: coping occurs seldom and when it happens, coping is not always overtly manifested. Therefore, in order to identify the cop- ing strategies employed by users in situ longitudinal observa- tions have to be conducted, which are resource intensive. We propose a more economical method that transfers the coping strategies employed by groups of users that cope frequently and overtly, such as people with disabilities, to broader populations. To do so, we first identify the coping strategies employed by people with disabilities; then we code these strategies and convert them into coping detection algorithms that are injected into web pages. Remote longitudinal studies are run with broader populations to measure the detection rate of the algorithms. Based on participants’ feedback we iteratively modify the algorithms to adjust them to the coping strategies users employ. We illustrate this method with a case study that transfers the strategies employed by visually disabled users to able-bodied users. We discover that different populations do not only face the same problems, but also exhibit similar strategies to tackle them.

当用户在网络中遭遇困境时，会采用极端适配手段来应对问题情境，这类手段即称为应对策略（coping strategies）。若能自动识别此类情境，我们便可提供绕过或预先规避这些问题的解决方案。然而，提取此类应对策略是一项极具挑战性的任务：这类行为发生频率极低，且即便出现，也并非总是外显可察的。因此，为识别用户在实际使用场景中采用的应对策略，需开展实地纵向观测，而这类观测的资源消耗极高。为此，我们提出一种更经济高效的方法：将频繁且外显地采用应对策略的用户群体（如残障人士）所使用的应对策略，迁移推广至更广泛的用户群体。具体而言，我们首先识别残障人士所采用的应对策略；随后对这些策略进行编码，并将其转化为可嵌入网页的应对策略检测算法。我们针对更广泛的用户群体开展远程纵向研究，以评估这些算法的检测准确率。基于参与者的反馈，我们对算法进行迭代优化，使其适配不同用户实际采用的应对策略。我们通过一项案例研究验证该方法：将视障用户所采用的应对策略迁移至健全用户群体。研究发现，不同用户群体不仅会遭遇同类问题，且解决问题的应对策略也具有相似性。