Automated Reliable Reasoning and Verification Using Neurosymbolic Approaches for Software Engineering

Software engineering has long relied on symbolic techniques (such as formal logic, program analysis, knowledge representation, and formal verification) to establish software properties. In parallel, machine learning has enabled systems to learn from code, specifications, execution traces, and operational data. However, learned models often lack explicit semantics, transparent reasoning, and verifiable guarantees, limiting their use in high-assurance settings. Neurosymbolic approaches combine data-driven learning with symbolic knowledge and reasoning, enabling systems that can both learn and provide structured explanations or evidence for their behavior. This integration offers opportunities to automate software engineering tasks while improving correctness, traceability, and interpretability. At the same time, it raises important questions regarding the representation of software knowledge, the interaction of neural and symbolic components, and the evaluation, verification, and assurance of software systems whose behavior depends on both learned and explicitly specified elements.This Research Topic aims to advance theories, methods, architectures, and engineering practices for reliable automated reasoning and verification using neurosymbolic approaches in software engineering. We define neurosymbolic integration as a substantive interaction between neural and symbolic representations, learning processes, or reasoning mechanisms. Examples include learning under formal constraints, neural models operating on symbolic knowledge structures, differentiable reasoning, and architectures in which neural and symbolic components jointly influence system behavior. Purely symbolic systems, purely neural systems, or approaches that use symbolic methods only as post-hoc validation generally fall outside the scope of the collection.We seek contributions that demonstrate how such integration can improve reasoning, verification, correctness, safety, robustness, explainability, accountability, or other software engineering outcomes. Topics of interest include the integration of formal specifications, program semantics, theorem proving, knowledge graphs, causal models, and probabilistic logic with learning-based systems. The Research Topic also aims to identify appropriate assurance criteria, benchmarks, engineering patterns, and evaluation methodologies. Contributions should support claims through appropriate theoretical, formal, empirical, benchmark-based, or practical evidence and help establish a systematic understanding of reliable neurosymbolic software systems.We invite original research articles, reviews, methods, tools, benchmarks, case studies, experience reports, and perspective papers on neurosymbolic software engineering. Relevant topics include:- neurosymbolic architectures and design patterns;- automated reasoning, theorem proving, constraint solving, and formal verification;- validation, testing, runtime assurance, and certification;- program synthesis, repair, and analysis;- knowledge representation, ontologies, causal reasoning, and knowledge graphs;- neurosymbolic agents and multi-agent systems;- learning under logical or semantic constraints;- formal foundations of hybrid systems;- explainability, robustness, safety, and trustworthy AI;- benchmarks, datasets, metrics, and reproducible evaluation frameworks.Submissions should clearly identify the neural and symbolic components, explain how they interact, and demonstrate their contribution to software engineering outcomes. We welcome both positive and negative findings, including replication studies, scalability analyses, failure-mode investigations, boundary-finding studies, and other work that improves understanding of the capabilities and limitations of neurosymbolic approaches. Topic editors: Ruben Ruiz-Torrubiano, Sona Ghahremani Source: https://www.frontiersin.org/research-topics/84751/automated-reliable-reasoning-and-verification-using-neurosymbolic-approaches-for-software-engineering