Synthesis of attributed feature models from product descriptions

Abstract

Many real-world product lines are only represented as nonhierarchical collections of distinct products, described by their configuration values. As the manual preparation of feature models is a tedious and labour-intensive activity, some techniques have been proposed to automatically generate boolean feature models from product descriptions. However, none of these techniques is capable of synthesizing feature attributes and relations among attributes, despite the huge relevance of attributes for documenting software product lines. In this paper, we introduce for the first time an algorithmic and parametrizable approach for computing a legal and appropriate hierarchy of features, including feature groups, typed feature attributes, domain values and relations among these attributes. We have performed an empirical evaluation by using both randomized configuration matrices and real-world examples. The initial results of our evaluation show that our approach can scale up to matrices containing 2,000 attributed features, and 200,000 distinct configurations in a couple of minutes. CCS Concepts •Software and its engineering → Software product lines; •Social and professional topics → Software reverse engineering;

Automated extraction of product comparison matrices from informal product descriptions

“…First, a PCM is per se a widely used abstraction for understanding and comparing products within a domain [53,7,8]. Second, the formation of feature models from a PCM is possible with synthesis techniques [10,7,6,11,12,52]. However, as shown in our empirical study, extracted …”

“…It is then immediate to identify recurrent features of a domain, to understand the specific characteristics of a given product, or to locate the features supported and unsupported by some products. PCMs are also an interesting potential step stone for further analysis such as: (1) formalization and generation of other domain models (e.g., feature models [10,7,11,12,13]), (2) feature recommendation [6], (3) automatic reasoning (e.g., multi-objective optimizations) [14], (4) derivation of automatic comparators and/or configurators [9].…”

Using machine learning to infer constraints for product lines

“…These works are not applicable in our context since i) we have to learn from a set of faulty configurations (not from textual content); ii) we have to deal with constraints among numerical values. Bécan et al [6] targeted the problem of synthesizing attributed feature models, including constraints among attributes and features. We also produce such kinds of constraints, but we do not assume that the original set of valid configurations is sound and complete.…”

A complexity tale

“…Existing works address the synthesis of FMs from both product specifications [5,6,18,24,31,32] and collections of textual documents [19,20,23,34].…”

“…Then satisfiability techniques are applied on the formula to compute variability information. As many different FMs can be built from the same formula, the extraction requires heuristics for guiding the selection of the proper organization of features (i.e., feature hierarchy, feature groups, attribute placement) [5,6].…”