一种融合视觉决策理论预测用户选择的方法

为有效解决在眼动人机交互环境中由于难以判断用户决策而导致的米达斯接触问题, 结合视觉决策理论, 通过融合眼动指标与时间序列来预测用户行为。先记录用户浏览目标页面时的眼动数据及进行选择时的点击数据, 并对数据按时间轴进行采样, 得到注视序列; 然后根据注视序列拟合用户进行决策时的注视似然曲线, 确定相关参数与触发阈值。眼动实验结果表明, 单凭注视阈值预测准确率就可以达到78. 3%, 在人机交互环境中拟合的似然曲线与视觉决策理论耦合。方法的实施无须用户进行额外的配合, 不依赖预设的固定阈值; 相对于设置固定触发阈值, 或要用户进行眨眼、按键等操作的传统方法, 交互过程可更顺畅、自然。

Visual decision making theory combinedmethod to predict user choice

In order to overcome Midas touch problem in eye movement-based human-computer interaction dialog, which was caused by lack of knowledge to user decision processes, this paper proposed a method which combined with visual decision making theory, to predicate user decision through syncretism analysis of eye movements indices and time sequence. First, it needed to collect eye movements data and click through data in target page, and sample data with timeline, obtained the fixation array; then fitted the gaze likelihood curve with fixation array, confirmed related parameter and trigger threshold. Results of the eye tracking experiments show that, the likelihood curve derives from human-Web interaction environment coupling to visual decision making theory; the prediction rate purely based on fixation threshold reaches 78. 3%. The implement of this method do not need additional coordination from users, do not rely on stiff threshold; contract to traditional methods of setting settled threshold, or doing operations such as blinking and key press, the interaction process could be more influence and nature.