Probabilistic software product lines

Abstract

We introduce a probabilistic extension of our previous work SPLA: a formal framework to specify and analyze software product lines. We use probabilistic information to identify those features that are more frequently used. This is done by computing the probability of having a feature in a specific software product line, from now on SPLAP . We redefine the syntax of SPLA to include probabilistic operators and define new operational and denotational semantics. We prove that the expected equivalence between these two semantic frameworks holds. Our probabilistic framework is supported by a set of scripts to show the model behavior. We briefly comment on the characteristics of the scripts and discuss the advantages of using probabilities to quantify the likelihood of having features in potential software product lines.