Requirements modeling languages for software product lines: A systematic literature review

• There is a concern for generating proposals with higher levels of expressiveness.
• There is not a strong relationship between the proposals and SPL development process.
• There is a need for better ways to validate the modeling proposals.
• The proposals have a low level of empirical validation and adoption in industry.
• The level of maturity, expressive power and tool support of the proposals is low.

Context: Software product lines (SPLs) have reached a considerable level of adoption in the software industry, having demonstrated their cost-effectiveness for developing higher quality products with lower costs. For this reason, in the last years the requirements engineering community has devoted much effort to the development of a myriad of requirements modelling languages for SPLs.Objective: In this paper, we review and synthesize the current state of research of requirements modelling languages used in SPLs with respect to their degree of empirical validation, origin and context of use, level of expressiveness, maturity, and industry adoption.Method: We have conducted a systematic literature review with six research questions that cover the main objective. It includes 54 studies, published from 2000 to 2013.Results: The mean level of maturity of the modelling languages is 2.59 over 5, with 46% of them falling within level 2 or below -no implemented abstract syntax reported-. They show a level of expressiveness of 0.7 over 1.0. Some constructs (feature, mandatory, optional, alternative, exclude and require) are present in all the languages, while others (cardinality, attribute, constraint and label) are less common. Only 6% of the languages have been empirically validated, 41% report some kind of industry adoption and 71% of the languages are independent from any development process. Last but not least, 57% of the languages have been proposed by the academia, while 43% have been the result of a joint effort between academia and industry.Conclusions: Research on requirements modeling languages for SPLs has generated a myriad of languages that differ in the set of constructs provided to express SPL requirements. Their general lack of empirical validation and adoption in industry, together with their differences in maturity, draws the picture of a discipline that still needs to evolve.

Figure optionsDownload as PowerPoint slide