User Interface Reverse Engineering in Support of Interface Migration to the Web

• Published in: Automated software engineering Vol. 10; no. 3; pp. 271 - 301
• Main Authors: Stroulia, E, El-Ramly, M, Iglinski, P, Sorenson, P
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Nature B.V 01.07.2003
• Subjects: Accessibility; Reverse engineering; Screens; User interface; User interfaces
• ISSN: 0928-8910; 1573-7535
• DOI: 10.1023/A:1024460315173

Legacy systems constitute valuable assets to the organizations that own them, and today, there is an increased demand to make them accessible through the World Wide Web to support e-commerce activities. As a result, the problem of legacy-interface migration is becoming very important. In the context of the CELLEST project, we have developed a new process for migrating legacy user interfaces to web-accessible platforms. Instead of analyzing the application code to extract a model of its structure, the CELLEST process analyzes traces of the system-user interaction to model the behavior of the application's user interface. The produced state-transition model specifies the unique legacy-interface screens (as states) and the possible commands leading from one screen to another (as transitions between the states). The interface screens are identified as clusters of similar-in-appearance snapshots in the recorded trace. Next, the syntax of each transition command is extracted as the pattern shared by all the transition instances found in the trace. This user-interface model is used as the basis for constructing models of the tasks performed by the legacy-application users; these task models are subsequently used to develop new web-accessible interface front ends for executing these tasks. In this paper, we discuss the CELLEST method for reverse engineering a state-transition model of the legacy interface, we illustrate it with examples, we discuss the results of our experimentation with it, and we discuss how this model can be used to support the development of new interface front ends.