基于视觉行为的健康信息甄别中任务难度感知变化预测研究

Prediction of Perceived Changes in Task Difficulty in Health Information Identification Based on Visual Behavior

[目的/意义]任务难度感知反映了用户健康信息甄别过程中的认知状态。预测用户预期难度到体验难度的感知变化,有助于健康信息平台服务方和管理方为用户提供针对性的服务,提高用户甄别效率及体验。[方法/过程]通过受控实验收集用户健康信息甄别过程中查询、浏览和查看详情三类基本活动下的注视和扫视行为数据,并分别分析预期难度和体验难度下用户视觉行为差异,获取能表征用户感知难易的视觉行为指标,进而借助机器学习方法建立两阶段预测模型实现从预期难度到体验难度感知变化的预测。[结果/结论]健康信息甄别中用户预期难度在查询和浏览活动下的注视行为具有显著差异,而用户体验难度在三类活动下的视觉行为均具有显著差异,据此构建的预测模型方案能够有效预测用户任务难度感知的变化,且对用户任务难度感知增加或持续易时的预测效果最优,准确率超过90%。

[Purpose/significance]Task difficulty perception reflects the cognitive state of users in the process of health information identification.Predicting the perceived change from expected difficulty to experienced difficulty of users helps health information platform service providers and administrators provide targeted services to users and improve their identification efficiency and experience.[Method/process]Through controlled experiments,we collected data on users’fixation and saccade behavior under the three basic activities of querying,browsing and viewing details in the process of health information identification,analyzed the differences in users’visual behavior under the expected difficulty and experienced difficulty respectively,obtained visual behavior indicators that can characterize users’perceived difficulty,and then established a two-stage prediction model with the help of machine learning methods to realize the prediction of perceived changes in task difficulty.[Result/conclusion]Under expected difficulty,the gaze behaviors under query and browsing activities have significant differences,while under experience difficulty,the visual behaviors under all three types of activities have significant differences.Accordingly,the prediction model scheme constructed in this paper can effectively predict the changes of user task difficulty perception,and the prediction effect is optimal when the user task difficulty perception increases or continues to be easy,with an accuracy rate of over 90%.