Empirical Comparison of Comprehensibility of Requirement Specification Techniques based on Natural Languages and Activity Diagrams

Abstract

Understandability belongs to the most important features of good quality software requirement specification (SRS). There exist plenty notations used for defining SRS, but still natural language (NL) belongs to the most popular. The specification written is NL could suffer from ambiguity, however it can be read by everybody without specific training. To eliminate, even partially, the drawbacks mentioned previously, SRS is written according to well defined guidelines and with the use of templates, e.g. use-case model consisting of a use-case diagram with a set of use-case detailed descriptions. Use-case descriptions are defined in NL or with dynamic diagrams, e.g. activity diagrams. This paper presents a controlled experiment which aimed at comparison of comprehensibility of techniques based on natural language and activity diagrams. The results of the experiment confirmed that formal notation is less ambiguous. Additionally, if a reader is accustomed to it, reading activity diagram not necessarily is time consuming.