Uncovering product line variability from early requirement documents

Mass production of customer-specific software application through software product lines has been gaining great attention in the past years. A software product line supports fast production of customized software applications by the composition of variable requirements, namely variability. Practitioners and researchers suggest that the efficient construction of software product lines depends on the ability of domain engineers to early identify potential variability. Controversially, uncovering product line variability from elicited requirements remains one of the main challenges in domain engineering. The current practice is an ad-hoc, tacit and consequently error-prone identification of variable requirements by domain experts while reviewing different versions of specification documents for similar products. Therefore, variability uncovering could represent an adoption barrier for many companies that should otherwise benefit. To cope with this challenge on product line requirement engineering, we propose in this paper a novel technique for uncovering variability from early requirement documents, specially, from existing Language Extended Lexicons (LEL). The technique suggests the analysis of LEL following a set of heuristics, which therefore, supports the precise grouping, identification and relation of potential variable requirements. In this paper we also illustrate the proposed technique through examples for the meeting scheduler domain.