Perceptual Dependencies in Information Visualization Assessed by Complex Visual Search
Perceptual Dependencies in Information Visualization Assessed by Complex Visual Search
复制标题
DOI：
10.1145/1278760.1278763
复制
发表时间：
2008-01-01
影响因子：
1.6
通讯作者：
Roerdink, Jos B. T. M.
中科院分区：
文献类型：
--
作者：
Van den Berg, Ronald;Cornelissen, Frans W.;Roerdink, Jos B. T. M.
可视化数据集的常见方法是将它们映射到图像，其中不同的数据维度映射到不同的视觉特征，例如颜色、大小和方向。在这里，我们考虑不同数据维度应受到同等重视和关注的可视化。对这些图像执行的许多最终用户任务都涉及某种形式的视觉搜索。通常，简单地假设在此类任务中可以相互独立地判断特征。然而，有证据表明当同时呈现多个特征时存在感知依赖性。这种依赖性可能会影响包含用于编码信息的特征组合的信息可视化，从而使受试者对不同数据维度的相关性进行不平等的加权。我们通过实验评估（1）可视化任务中判断依赖性的存在（在节点链接图中搜索目标节点）以及（2）特征对比度与显着性的关系。从可视化的角度来看，我们最相关的发现是（a）为了均衡大小和颜色的显着性（以及自下而上的权重），颜色对比度必须变得非常低。此外，方向不太适合表示由大范围数据值组成的信息，因为它没有显示对比度和显着性之间的明确关系； (b) 颜色和大小是可以独立用来表示信息的特征，至少就我们研究中使用的颜色范围而言； (c)（静态）特征显着层次结构的概念是错误的；一个特征与另一个特征相比的显着程度不是固定的，而是特征对比的函数； (d) 最终决定似乎是与基于个人注视所获得的测量指标一样好的感知表现指标。因此，眼动追踪并不一定能为旨在评估搜索任务性能的用户研究带来好处。
A common approach for visualizing data sets is to map them to images in which distinct data dimensions are mapped to distinct visual features, such as color, size and orientation. Here, we consider visualizations in which different data dimensions should receive equal weight and attention. Many of the end-user tasks performed on these images involve a form of visual search. Often, it is simply assumed that features can be judged independently of each other in such tasks. However, there is evidence for perceptual dependencies when simultaneously presenting multiple features. Such dependencies could potentially affect information visualizations that contain combinations of features for encoding information and, thereby, bias subjects into unequally weighting the relevance of different data dimensions. We experimentally assess (1) the presence of judgment dependencies in a visualization task (searching for a target node in a node-link diagram) and (2) how feature contrast relates to salience. From a visualization point of view, our most relevant findings are that (a) to equalize saliency (and thus bottom-up weighting) of size and color, color contrasts have to become very low. Moreover, orientation is less suitable for representing information that consists of a large range of data values, because it does not show a clear relationship between contrast and salience; (b) color and size are features that can be used independently to represent information, at least as far as the range of colors that were used in our study are concerned; (c) the concept of (static) feature salience hierarchies is wrong; how salient a feature is compared to another is not fixed, but a function of feature contrasts; (d) final decisions appear to be as good an indicator of perceptual performance as indicators based on measures obtained from individual fixations. Eye tracking, therefore, does not necessarily present a benefit for user studies that aim at evaluating performance in search tasks.