Akademik Veri Yönetim Sistemi
9th ACM SIGCHI Symposium on Engineering Interactive Computing Systems, EICS 2017, Lisbon, Portekiz, 26 - 29 Haziran 2017, ss.3-8, (Tam Metin Bildiri)
Web-based interactive systems can be verified and validated with eye tracking which provides valuable insights for understanding user interactions. However, individual eye movements (scanpaths) tend to be complicated and different from each other. Existing algorithms focus on clustering multiple scanpaths into a single representative path, but do not tolerate differences. This may result in excluding some popular elements from the resulting path and the representativeness of the path can decrease. Scanpath Trend Analysis (STA) is one of these algorithms which is being increasingly used. It can be used for identifying trending areas of interest (AoIs) on a web page among multiple users and discovering their trending path in terms of these AoIs. Compared to the outputs of other algorithms, the output of the STA algorithm is the most representative. Since this algorithm considers all given scanpaths while discovering their trending path, it can be negatively affected when the variance is high between the scanpaths. In this paper, we introduce a new parameter to the STA algorithm, called tolerance level. This parameter makes the algorithm more tolerant by allowing the identification of trending AoIs based on a subset of given scanpaths instead of all of them. Our study shows that the STA algorithm can achieve more representative path with a tolerance. Thus, the validity and verification of interactive systems can be investigated with more representative path.