Pinpointing User Interface Deficiencies Using Pattern Recognition Techniques
|dc.contributor.author||Dasari Kali Venkata, Divya ( )|
|dc.identifier.citation||Dasari Kali Venkata, D. (2013). Pinpointing user interface deficiencies using pattern recognition techniques (Unpublished thesis). Texas State University, San Marcos, Texas.|
The Effort Based Model of usability aids in evaluating user interface (UI), development of usable software, and pinpointing software usability defects. In this context, the term pinpoint analysis refers to identifying and locating software usability issues and correlating these issues with the UI software code. In this thesis, the underlying theories of the effort based model along with pattern recognition techniques are used to produce a framework for identifying usability deficiencies in software.
Often, when users are in a state of confusion and not sure how to proceed using the software, they tend to gaze around the screen trying to find the best way to complete a task. This behavior is referred to as excessive effort. In this work, pattern recognition techniques are applied to data gathered throughout user interaction with software in an attempt to identify excessive effort segments. This is done by logging all user activities as video and data files by an eye tracker. The data files are divided into segments using event based segmentation, where a segment is the time between two consecutive keyboard/mouse clicks. Subsequently, data reduction programs are run on the segments for generating feature vectors. Pattern recognition techniques like feature selection, thresholding, clustering, and principal component analysis (PCA) are applied to the features in order to automatically classify each segment into excessive and non-excessive effort segments. This allows developers to harness their effort and focus on the excessive effort segments that need attention.
To verify the results of the pattern recognition procedures, the video file is manually classified into excessive and non-excessive segments and the results of automatic and manual classification are compared. Experiment results show more than 40% reduction in time for usability testing. Of all the methods used, experiments using the threshold method using the number of fixations and a threshold method applied to the first principal component produce good results which are significantly better than results obtained through other experiments.
|dc.format.medium||1 file (.pdf)|
|dc.subject||Pattern recognition techniques|
|dc.title||Pinpointing User Interface Deficiencies Using Pattern Recognition Techniques|
|thesis.degree.grantor||Texas State University||en_US|
|thesis.degree.name||Master of Science||en_US|