Improving Sequence Diagram Modeling Performance

Abstract

The Unified Modeling Language (UML) has become the de facto standard for object-oriented software development. It has been widely adopted in both training and practice. However, UML has often been criticized for being overly complex and difficult to learn for novice analysts. Although some research studies have identified specific novice difficulties in learning UML, there is little research proposing viable techniques for addressing these difficulties. In particular, there is a lack of research evaluating the usability of the sequence diagram (SD), which models the interactions among objects of a software application. This paper reports a research study that proposes a technique called “CHOP” (CHunking, Ordering, Patterning), which is designed to improve novice analyst performance in modeling an SD. The CHOP technique is based on the Cognitive Load Theory (CLT) and was developed by addressing the three types of cognitive load encountered by novices. An experimental study testing the efficacy of the CHOP technique in comparison to the worked-example approach indicated that the CHOP technique significantly improves novice analyst’s ability to model interactions among objects; however, the worked-example technique was the more efficient during training. The study also found that subjects using the CHOP technique rated its perceived usefulness significantly higher than subjects using the worked-example approach.