UCWE: A User-Centric Approach for Web Quality of Experience Measurement
Xiaohui Zheng, Yan Jin, Xiao Shi, Yonghao Song, Xiaofang Zhao · 2019
Quality of Experience (QoE) is a key factor that intuitively re-flects users' satisfaction and perception of web services. How-ever, due to the nature of its subjectivity, it is difficult to be measured and quantified accurately. In fact, we argue that QoE, which has multidimensional metrics, should be measured from a comprehensive perspective. To achieve this, a novel tool that is used to collect objective and subjective data, Data Col-lecting with Crowdsourcing (DCC), is presented and corre-spondingly, a hybrid measurement approach User-Centric Web QoE Evaluating (UCWE) is proposed. This approach quantifies QoE based on subjective ratings which are directly appraised by users and objective metrics which are gathered during the rating process. Based on the collected data, UCWE is able to analyze user-based metrics and build an accurate and adaptive Web QoE evaluating model. In evaluation, training data from over 5000 web accesses are used. Experimental re-sults show that our approach achieves higher accuracy than existing well-known Machine Learning (ML) methods by about 5%. Besides, different groups of users have different rating preferences and effects on QoE.