Visual Evaluation Tool for Eye-Tracking Data
In eye-tracking studies, usually there are problems in defining the method of data analysis. You can use ANOVA, different clustering, Markov fields, but the question about what exactly we should measure (and for what) and which method is better is still actual.
Visual Evaluation Tool (VETool) was developed for visual evaluation of eye-tracking data and renders. The idea to develop such kind of tool was came up in the Workshop at the 13th KOLI CALL INTERNATIONAL CONFERENCE ON COMPUTING EDUCATION RESEARCH.
VETool is build on the NetBeans Platform in Java and JavaFX. Why NetBeans? Hmm...
Because it is cool and free. :) VETool is open source under GLP 2: https://bitbucket.org/orlovpa/visual-evaluation-tool-vetool
The goal of the workshop was to describe the reading process of source code in Java. By analysing gaze data during code reading processes, eye-movement patterns were defined, which are an essential part of programming comprehension.
I decided to combine two data types for visualization of source code reading. First, is the data from ELAN software (from Max Planck Institute). From ELAN we get information about the duration of seeing in different source code blocks. We need to annotate by hand the gaze fixation, for example, if the programmer sees the constructor we need to mark the time of the starting point, as well as the end time. Finally, we get quite a big list of annotations separated by blocks and sub blocks, patterns of reading, and strategies.
Visual Evaluation Tool (VETool) can show you more than I am able to explain:
Here you can see that two programmers looked at the constructor at the beginning of task solving and at the middle. But, also you can see that there is no strong significance here. In the right panel there are other possible blocks of source code for visualisation, such as methods, definitions, etc.
The second kind of the data is gaze fixation duration data from Ogama software (OpenGazeAndMouseAnalyzer).
A full screen demo is here:
I hope to provide more features in the next release and feedback is, of course, very welcome.
(Note: Opinions expressed in this article and its replies are the opinions of their respective authors and not those of DZone, Inc.)