权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Not just 'how', but 'why' - Exploring the effects of explainability on software learnability

不仅仅是“如何”，而是“为什么”——探索可解释性对软件可学习性的影响

基本信息

批准号：
2715760
负责人：
金额：
--
依托单位：
University College London
依托单位国家：
英国
项目类别：
Studentship
财政年份：
2022
资助国家：
英国
起止时间：
2022 至无数据
项目状态：
未结题

来源：
https://gtr.ukri.org/projects?ref=studentship-2715760
关键词：
just how but why Exploring

项目摘要

Tutorials are tools that help users learn about a software. A number of different approaches can be adopted in the design of a tutorial. Some examples of popular approaches include, and can sometimes involve a combination of: step-by-step instructions, demonstrations, gamification, and exploratory learning. However, current approaches in tutorial design tend to focus on memorisation strategies, which may be insufficient in supporting user learning for complex software. Helping users to develop accurate mental models may address this issue. Research from the domain of Explainable Artificial Intelligence (XAI) may provide relevant directions for investigation. This proposal suggests exploring the way in which explainability could affect software learnability. This will be done via a digital variation of the question-asking protocol, comparing the effects of four different tutorials (control, baseline tutorial, explainability features only, and explainable tutorial) on user learning and performance. Methods used will be mostly quantitative, since the primary aim of this project is to investigate the effects of explainability on software learnability, exploring measures such as time spent exploring other resources and numbers of queries. However, some qualitiative methods will be used for the testing the final variable, 'software learnability', because it will be important to assess the types of queries asked, providing greater insight into the types of information users believe they are missing.

教程是帮助用户了解软件的工具。在教程的设计中可以采用许多不同的方法。流行方法的一些例子包括，有时可能涉及以下几种方法的组合：逐步指导，演示，游戏化和探索性学习。然而，目前的教程设计方法往往侧重于记忆策略，这可能不足以支持用户学习复杂的软件。帮助用户开发准确的心理模型可以解决这个问题。可解释人工智能（XAI）领域的研究可能会为调查提供相关方向。这个提议建议探索可解释性影响软件可学习性的方式。这将通过提问协议的数字变体来完成，比较四种不同教程（对照、基线教程、仅可解释功能和可解释教程）对用户学习和性能的影响。所使用的方法将主要是定量的，因为这个项目的主要目的是调查软件可学习性的可解释性的影响，探索措施，如花费在探索其他资源和查询的数量的时间。然而，一些定性方法将用于测试最终变量“软件可学习性”，因为评估所问问题的类型非常重要，可以更深入地了解用户认为他们遗漏的信息类型。