Visual Similarity Perception of Directed Acyclic Graphs: A Study on Influencing Factors and Similarity Judgment Strategies
Visual Similarity Perception of Directed Acyclic Graphs: A Study on Influencing Factors and Similarity Judgment Strategies
复制标题
DOI：
10.7155/jgaa.00467
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
K. Ballweg;M. Pohl;Günter Wallner;T. V. Landesberger
中科院分区：
文献类型：
--
作者：
K. Ballweg;M. Pohl;Günter Wallner;T. V. Landesberger
有向无环图（dag）的视觉比较通常在各个学科（例如，金融，生物学）中遇到。尽管如此，关于人类对相似性感知的知识目前还是相当有限的。所谓相似性感知，是指人类如何感知dag的共性和差异性，并由此得出相似性判断。为了填补这一空白，我们努力确定影响DAG相似性感知的因素。因此，我们采用定量和定性分析方法进行了卡片分类研究，以确定(1)参与者认为相似的dag组和(2)分组背后的原因。我们还对收集到的数据进行了扩展分析，以(1)揭示影响因素的具体情况，(2)调查采用哪些策略来得出相似度判断。我们的研究结果表明，DAG相似度感知主要受层次数、层次上节点数和DAG整体形状的影响。我们还确定了参与者用于形成相似dag组的三种策略：分而治之，尊重整个数据集并逐一考虑因素，以及考虑单个因素。具体的因素是，例如，人类在判断dag的相似性时平均考虑四个因素。建立对这些过程的理解可以告知比较可视化的设计和与它们交互的策略。交互策略必须允许用户将其相似性判断策略应用于数据。所考虑的因素包含信息，例如，哪些因素被人类忽视，因此需要通过可视化来突出显示。
Visual comparison of directed acyclic graphs (DAGs) is commonly encountered in various disciplines (e.g., finance, biology). Still, knowledge about humans’ perception of their similarity is currently quite limited. By similarity perception, we mean how humans perceive commonalities and differences of DAGs and herewith come to a similarity judgment. To fill this gap, we strive to identify factors influencing the DAG similarity perception. Therefore, we conducted a card sorting study employing a quantitative and qualitative analysis approach to identify (1) groups of DAGs the participants perceived as similar and (2) the reasons behind their groupings. We also did an extended analysis of our collected data to (1) reveal specifics of the influencing factors and (2) investigate which strategies are employed to come to a similarity judgment. Our results suggest that DAG similarity perception is mainly influenced by the number of levels, the number of nodes on a level, and the overall shape of the DAG. We also identified three strategies used by the participants to form groups of similar DAGs: divide and conquer, respecting the entire dataset and considering the factors one after the other, and considering a single factor. Factor specifics are, e.g., that humans on average consider four factors while judging the similarity of DAGs. Building an understanding of these processes may inform the design of comparative visualizations and strategies for interacting with them. The interaction strategies must allow the user to apply her similarity judgment strategy to the data. The considered factors bear information on, e.g., which factors are overlooked by humans and thus need to be highlighted by the visualization.