Verification of Hierarchical Artifact Systems

Abstract

Data-driven workflows, of which IBM's Business Artifacts are a prime exponent, have been successfully deployed in practice, adopted in industrial standards, and have spawned a rich body of research in academia, focused primarily on static analysis. The present work represents a significant advance on the problem of artifact verification, by considering a much richer and more realistic model than in previous work, incorporating core elements of IBM's successful Guard-Stage-Milestone model. In particular, the model features task hierarchy, concurrency, and richer artifact data. It also allows database key and foreign key dependencies, as well as arithmetic constraints. The results show decidability of verification and establish its complexity, making use of novel techniques including a hierarchy of Vector Addition Systems and a variant of quantifier elimination tailored to our context.

Formal Modeling and SMT-Based Parameterized Verification of Data-Aware BPMN

“…From the foundational perspective we are interested in equipping DABs with datatypes and corresponding rigid predicates, including arithmetic operators, as done in [14] for artifact systems. This is promising especially considering that there are plenty of state-of-the-art SMT techniques to handle arithmetics.…”

“…Monotonic abstractions could introduce spurious unsafe traces, and in fact MCMT warns the user about this (in practice, not so frequent) possibility. An orthogonal, challenging question is how, and to what extent, some of the most recent techniques developed for temporal model checking of artifact-centric systems [14] can be incorporated in our approach, allowing us to prove more sophisticated properties beyond safety.…”

“…We recall the key notions behind array-based systems, and the array-based artifact systems recently studied in [5] to bridge the gap between SMT-based model checking of array-based systems [17,18], and verification of data-and artifact-centric processes [12,14].…”

“…This triggered a flourishing research on the formalization and the boundaries of verifiability of data-aware processes, focusing mainly on data-and artifact-centric models [2,12]. Recent results in this stream of research [14,5] come with two strong advantages. First, they consider the relevant setting where the running process evolves a set of relations (henceforth called a data repository) containing data objects that may have been injected from the external environment (e.g., due to user interaction), or borrowed from a read-only relational database with constraints (henceforth called catalog).…”

SMT-based verification of data-aware processes: a model-theoretic approach

“…We call artifact systems of this form Simple Artifact System (SAS ). Instead, more sophisticated types of artifact systems have been studied recently in Deutsch et al (2016), Li et al (2017), Deutsch et al (2019). Here, the working memory is not only equipped with artifact variables as in SAS , but also with so-called artifact relations, which supports storing arbitrarily many tuples, each accounting for a different artifact instance that can be separately evolved on its own.…”