Capturing Software Traceability Links from Developers' Eye Gazes

Braden Walters*, Timothy Shaffer**, Bonita Sharif*, Huzefa Kagdi***
*Department of Computer Science and Information Systems Youngstown State University Youngstown, Ohio 44555 USA
**Department of Mathematics and Statistics Youngstown State University Youngstown, Ohio 44555 USA
***Department of Electrical Engineering and Computer Science Wichita State University Wichita, Kansas 67260 USA
bmwalters01@student.ysu.edu, trshaffer@student.ysu.edu, bsharif@ysu.edu, kagdi@cs.wichita.edu

A snapshot of iTrace's working environment is shown below

iTrace snapshot

Subject System
iTrust study environment used with injected bugs.

Algorithm implementation
SimpleGraph Gaze-Link algorithm implementation (on github)

Links generated by SimpleGraph Gaze-Link for each task:
True Links
The true links for each task

Background Questionnaire

Study Material - Stage One

Study Material - Stage Two

Some Graphs of Linked Entities (optional output of the SimpleGraph Gaze-Link algorithm)

The graphs show the visual path of exploration. Entities that are placed closer together have stronger links between them and in turn with the task at hand.

Three graphs are shown below for task 1, 4, and 5 respectively. The graph for task 4 is an example of a very direct graph where the developer knew exactly where to find the answer. The graphs for Task 1 and task 5 have more interesting shapes. Task 1's graph (and Task 4's graph) shows a linear progression. Task 5's graph shows two different paths taken by the developer to solve the problem.

Task 1 - one subject's graph


Task 4 - one subject's graph


Task 5 - one subject's graph



Email bsharif@ysu.edu for further information