Graph Construction and Matching for Imperative Programs using Neural and Structural Methods

2026-04-29 · Source: cs.SE updates on arXiv.org · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

A new pipeline for "Graph Construction and Matching for Imperative Programs" converts imperative programs and their annotations into typed, attributed graphs. Developed by Beg, O'Donoghue, and Monahan, this 14-page work, submitted on April 29, 2026, and revised on July 12, 2026 (v3), aims to facilitate the reuse of verification artifacts by identifying structural and semantic similarities across programs and specifications. The pipeline integrates abstract syntax tree parsing with semantic embeddings derived from models like SentenceTransformer and CodeBERT. Experiments, covering datasets such as C with ACSL, Java with JML, and Dafny programs, demonstrate its ability to construct consistent graph representations across different languages and annotation styles. This foundational work provides a practical basis for future semantic enrichment and approximate graph matching, enabling scalable verification artifact reuse.

Key takeaway

For verification engineers aiming to reuse program verification artifacts across different languages, this work offers a clear path. You should consider adopting graph-based representations for imperative programs and their annotations. This approach, using neural and structural methods, enables consistent cross-language analysis, reducing redundant verification efforts. Explore integrating AST parsing with semantic embeddings from models like CodeBERT to build scalable systems for artifact reuse.

Key insights

Converting programs to typed, attributed graphs using neural and structural methods enables scalable reuse of verification artifacts.

Principles

• Consistent graph representations span languages.
• Semantic embeddings enhance structural parsing.
• Graph matching supports verification reuse.

Method

The pipeline converts imperative programs and annotations into typed, attributed graphs by integrating AST parsing with semantic embeddings from models like SentenceTransformer and CodeBERT.

In practice

• Apply to C with ACSL, Java with JML, Dafny.
• Use SentenceTransformer or CodeBERT for embeddings.
• Facilitate cross-language verification artifact reuse.

Topics

• Graph Construction
• Program Verification
• Semantic Embeddings
• Abstract Syntax Trees
• CodeBERT
• SentenceTransformer
• Imperative Programs

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Geometry of Semantic Space: Comparative Study of Discrete and Continuous Models — Graph-based semantic models offer more interpretable meaning organization than transformer embeddings, despite similar local topology.
• Project-wise Comparison of Software Birthmarks Using Weighted Partial Similarity — Project-wise software birthmark comparison requires weighting module similarities and focusing on top matches to detect partial reuse and mitigate incidental similarity.
• EmbeddedKittens: An Evaluation of Code Embeddings for Scratch — Code embeddings, especially those capturing structural information, are effective for Scratch learning analytics, even without task-specific fine-tuning.
• TeamSLS at SemEval-2026 Task 13: Detecting Machine-Generated Code with CodeBERT and Structural Features — Combining semantic and structural code features improves AI-generated code detection but struggles with cross-language shifts.
• Königsberg at SemEval-2026 Task 13: Beyond Language Models: A Low-Resource Feature-Driven and Data-Flow Embedding Approach for Machine-Generated Code Detection — A hybrid framework efficiently detects machine-generated code using software metrics and frozen GraphCodeBERT embeddings, avoiding LLM fine-tuning overhead.
• VCG-Bench: Towards A Unified Visual-Centric Benchmark for Structured Generation and Editing — "Diagram-as-Code" with mxGraph XML enables precise, controllable generation and editing of structured visual content, surpassing pixel-based methods.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.SE updates on arXiv.org.