[Paper Review] Monitoring Constraints and Metaconstraints with Temporal Logics on Finite Traces

Giuseppe De Giacomo, Riccardo De Masellis+2 more|arXiv (Cornell University)|Apr 4, 2020

Business Process Modeling and Analysis4 citations

TL;DR

This paper proposes a formal and practical framework for monitoring linear-time temporal logic constraints and metaconstraints on finite traces using LDL_f, a powerful extension of LTL_f that captures monadic second-order logic over finite traces. By directly translating LDL_f formulae into nondeterministic finite-state automata and leveraging a four-valued runtime verification semantics (permanent/temporary satisfaction/violation), the approach enables declarative specification of both constraints and metaconstraints—enabling correct-by-construction monitoring via standard automata-theoretic techniques without ad-hoc algorithms.

ABSTRACT

Runtime monitoring is one of the central tasks in the area of operational decision support for business process management. In particular, it helps process executors to check on-the-fly whether a running process instance satisfies business constraints of interest, providing an immediate feedback when deviations occur. We study runtime monitoring of properties expressed in LTL on finite traces (LTLf), and in its extension LDLf. LDLf is a powerful logic that captures all monadic second order logic on finite traces, and that is obtained by combining regular expressions with LTLf, adopting the syntax of propositional dynamic logic (PDL). Interestingly, in spite of its greater expressivity, \LDLf has exactly the same computational complexity of LTLf. We show that LDLf is able to declaratively express, in the logic itself, not only the constraints to be monitored, but also the de-facto standard RV-LTL monitors. On the one hand, this enables us to directly employ the standard characterization of LDLf based on finite-state automata to monitor constraints in a fine-grained way. On the other hand, it provides the basis for declaratively expressing sophisticated metaconstraints that predicate on the monitoring state of other constraints, and to check them by relying on standard logical services instead of ad-hoc algorithms. In addition, we devise a direct translation of LDLf formulae into nondeterministic finite-state automata, avoiding to detour to Buchi automata or alternating automata. We then report on how this approach has been effectively implemented using Java to manipulate LDLf formulae and their corresponding monitors, and the well-known ProM process mining suite as underlying operational decision support infrastructure.

Motivation & Objective

• To address the lack of a formal, automata-theoretic foundation for runtime verification (RV) with four-valued semantics on finite traces.
• To extend existing LTL_f monitoring techniques to support metaconstraints that reason about the monitoring state of other constraints.
• To provide a direct, efficient translation of LDL_f formulae into nondeterministic finite-state automata, avoiding detours to Büchi or alternating automata.
• To implement the approach in a reusable Java library integrated with the ProM process mining framework for operational decision support.
• To lay the groundwork for extending the framework to data-aware monitoring with first-order temporal logics over finite traces.

Proposed method

• Define a four-valued runtime verification semantics for LDL_f that distinguishes permanent and temporary satisfaction/violation based on future trace continuations.
• Propose a direct translation of LDL_f formulae into nondeterministic finite-state automata (NFAs), preserving the semantics of the logic without relying on intermediate Büchi or alternating automata representations.
• Use the automata-theoretic characterization of LDL_f to compute reachability sets for each state, which determine the four truth values (perm_true, perm_false, temp_true, temp_false) in the RV semantics.
• Implement the monitoring process via an executable automaton that maintains a current state and updates it step-by-step upon event occurrences, returning the corresponding RV truth value after each transition.
• Leverage the Tweety library for model computation and the jautomata library for automata manipulation (union, intersection, determinization, trimming) to support formula analysis and monitor generation.
• Integrate the monitoring engine into the ProM process mining framework to enable real-time operational decision support in business process management.

Experimental results

Research questions

• RQ1 How can a formal, automata-theoretic foundation be established for four-valued runtime verification semantics on finite traces?
• RQ2 Can LDL_f be used to declaratively express not only constraints on process execution but also metaconstraints that reason about the monitoring state of other constraints?
• RQ3 What is an efficient and correct direct translation of LDL_f formulae into finite-state automata that supports the four-valued RV semantics?
• RQ4 How can the resulting monitors be effectively integrated into existing operational decision support systems like ProM?
• RQ5 What are the prospects for extending this framework to handle data-aware constraints expressed in first-order temporal logics over finite traces?

Key findings

• LDL_f, despite its greater expressivity (equivalent to monadic second-order logic on finite traces), has the same computational complexity as LTL_f, making it suitable for practical monitoring.
• The proposed direct translation of LDL_f into nondeterministic finite-state automata avoids the need for intermediate representations like Büchi or alternating automata, simplifying the monitoring pipeline.
• The four-valued runtime verification semantics—distinguishing permanent and temporary satisfaction/violation—can be correctly encoded in the automaton structure via reachability analysis from each state.
• The approach enables declarative specification of metaconstraints that predicate on the monitoring state of other constraints, such as 'if constraint φ is currently temporarily violated, then constraint ψ must be permanently satisfied'.
• The implementation in Java, integrated with ProM, demonstrates the feasibility and practicality of the approach for real-world business process monitoring.
• The framework provides a solid foundation for extending monitoring to data-aware settings, where events carry payloads and constraints reason over data values over time.