摘要: |
目的 利用随机森林决策树预测WEB交互界面跳动型视觉流程中视觉元素竞争结果的概率分布,为判断WEB界面用户视觉流合理性,以及优化WEB界面布局的用户视觉流提供依据。方法 讨论了界面跳动型视觉流的主要特点,提出WEB交互界面随机跳动的视觉流特点来源于其内在的视觉竞争机制,对于竞争的影响因子提出猜想,通过大量控制变量的视觉竞争实验验证猜想,得到因子与视觉竞争结果的相关性结论。对得到验证的影响因子,利用实验数据进行了随机森林算法训练,构建了模拟竞争过程的决策树模型,并利用模型对实际案例进行优化实验。结论 构建的随机森林决策树可以较好地模拟跳动型视觉流的视觉竞争机制。通过界面优化实验,验证该方法可以显著改善用户的视觉流程。 |
关键词: 随机森林算法 跳动型视觉流 决策树 视觉竞争 |
DOI:10.19554/j.cnki.1001-3563.2021.06.013 |
分类号:TB472 |
基金项目:2018年安徽省社会科学创新发展研究课题(2018CX023) |
|
Research on Optimization of Bounce Visual Flow Web Interface Based on Random Forest |
WEI Yan-li, LIU Yu-wei, ZHOU Xuan, LIU Han-yang
|
(HeFei University of Technology, Hefei 230000, China)
|
Abstract: |
The random forest decision tree is used to predict the probability distribution of the visual competition results of the WEB bounce visual flow interface, which provides a basis for determining the rationality of user visual flow on WEB interface and the user visual flow for optimizing the page layout. The main characteristics of the interface-bounce visual flow are discussed, and so the proposal that visual flow characteristics of the random jump of the WEB interactive interface are derived from its intrinsic visual competition mechanism is put forward. The conjectures for the competition influence factors is proposed and verified with a large number of control variables visual competition experiments, and the conclusion on the correlation between factors and visual competition results is obtained. For the varified impact factors, the experimental data is used to train the random forest algorithm, and the decision tree of the simulated competition process is constructed to carry out optimal experiment to actual cases. The constructed random forest decision tree can better simulate the visual competition mechanism of the bounce visual flow. Through the interface optimization experiment, it is verified that the method can significantly improve the user’s visual process. |
Key words: Graphics in Visual Communication Design[J]. Packaging Engineering, 2014, 35(8):9-13. |