学术论文

      基于特征推理的图标搜索特性实验研究

      Experimental study on characteristics of icon searching based on feature inference

      摘要:
      采用视觉搜索实验范式,通过行为和眼动追踪数据揭示了基于形状相似、功能关联和语义匹配特征3类推理条件的图标搜索特性,通过方差分析法比较了不同特征推理条件对各测量指标的影响.实验结果表明:形状相似性引导下的目标搜索时间以及对于各搜索项目的平均注视时间均最短,所消耗的认知资源最少,无需深度信息加工即可进行目标识别;基于功能关联性的图标搜索过程中注意资源分配最广;而语义词引导下的目标搜索正确率最高.基于形状相似性和功能关联性特征推理条件下的视觉搜索行为可简化为搜索项目与启动图标之间的模式匹配以及功能层级隶属关系判断过程;以语义词为启动特征的视觉搜索首先需经过语义词特征的图形化过程,再对搜索项进行匹配判断.3类搜索情境中均存在自上而下的信息加工特征.研究结果为人机界面中图标设计提供了理论指导.
      Abstract:
      Visual searching experimental paradigm was employed.Behavioral and eye-tracking data were recorded to reveal the characteristics of icon searching based on three types of feature inference conditions including visual similarities, functional relevance and semantic matching.Effects of different inference features on individual metric were compared through the analysis of variance.Experimental results show that time for target identification and mean fixation duration, which indicates the amount of cognitive resource allocated to each item, is the shortest when searching for the target icon according to shape similarity.No deep information processing is required to identify the target when based on the shape feature.The scope of attentional resources during icon searching is the broadest when based on functional relevance.The accuracy rate in the semantics-primed condition is the highest.Behaviors under guidance of shape similarity and functional revelance can be simplified as pattern matching or subordination judgment of functional hierarchy between each searching item and the priming icon.Semantic words-primed visual searching needs transformation of semantic word into image and then the image is matched with displayed items.Top-down information processing is included in these three searching situations.The research results provide theoritical guidance for icon design in digital interfaces.
      作者: 彭宁玥 薛澄岐
      Author: Peng Ningyue Xue Chengqi
      作者单位: 东南大学机械工程学院,南京,211189
      年,卷(期): 2017, 47(4)
      分类号: TB472 TP391.72
      在线出版日期: 2017年8月15日
      基金项目: 国家自然科学基金资助项目