Learning by sampling: learning behavioral family models from software product lines

Abstract

Family-based behavioral analysis operates on a single specification artifact, referred to as family model, annotated with feature constraints to express behavioral variability in terms of conditional states and transitions. Family-based behavioral modeling paves the way for efficient model-based analysis of software product lines. Family-based behavioral model learning incorporates feature model analysis and model learning principles to efficiently unify product models into a family model and integrate the behavior of various products into a behavioral family model. Albeit reasonably effective, the exhaustive analysis of product lines is often infeasible due to the potentially exponential number of valid configurations. In this paper, we first present a family-based behavioral model learning techniques, called FFSM Diff . Subsequently, we report on our experience on learning family models by employing product sampling. Using 105 products of six product lines expressed in terms of Mealy machines, we evaluate the precision of family models learned from products selected from different settings of the T-wise product sampling criterion. We show that product sampling can lead to models as precise as those learned by exhaustive analysis and hence, reduce the costs for family model learning.

A Benchmark for Active Learning of Variability-Intensive Systems

“…Family-based models are an enabler for such a structured analysis approach. Different types of family models have been used to analyze SPL functionality, including structural models [14,16] and behavioral models [5,9]. A type of structural model widely used in software product line engineering (SPLE) research is feature models [14,16].…”

“…A type of structural model widely used in software product line engineering (SPLE) research is feature models [14,16]. Structural models can be used to annotate traditional behavioral models to represent the behavior of a product line [5,9]. Featured Transition System (FTS) [5] and Featured Finite State Machines (FFSM) [10,11] are two examples of behavioral variability models for expressing the state-based behavior of families of software products.…”

“…Active model learning is a process in which a model of software behavior is learned by presenting different inputs and observing the output [1,8,20]. Model learning can be used in the behavioral analysis of an SPL, for example in model-based testing and model checking [9]. When applied to variability-intensive systems, it is expected that the interactions with different variants or products will help to identify commonalities and variability.…”

“…Adaptive model learning is a model learning approach in which previous models are reused to learn a new model, so that the speed of model learning can be increased [8]. In an SPL, the number of valid products may be exponential relative to the number of features [4,9]. However, due to the commonalities of SPL products, it is possible to reuse existing models during the model learning of an SPL.…”

“…However, due to the commonalities of SPL products, it is possible to reuse existing models during the model learning of an SPL. Therefore, it may be possible to reduce the total cost of learning the SPL model by using these common features [9].…”