Research map / Graph learning / Knowledge and temporal graphs
Learning to Represent Programs with Graphs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace the strongest sequence baseline with program graphs | VarMisuse · SeenProjTest | accuracy (%): 73.7 -> 85.5 | Helped |
| replace the strongest sequence baseline with program graphs | VarMisuse · UnseenProjTest | accuracy (%): 60.2 -> 78.2 | Helped |
| replace the strongest sequence baseline with program graphs | VarNaming · SeenProjTest | accuracy (%): 42.9 -> 53.6 | Helped |
| replace the strongest sequence baseline with program graphs | VarNaming · UnseenProjTest | F1 (%): 32.0 -> 62.0 | Helped |
| add graph edges beyond NextToken, Child, LastUse, and LastWrite (ablation) | Program Graph · SeenProjTest VarMisuse | accuracy (%): 80.6 -> 85.5 | Helped |
| add graph edges beyond NextToken, Child, LastUse, and LastWrite (ablation) | Program Graph · SeenProjTest VarNaming | accuracy (%): 31.2 -> 53.6 | Helped |
| add syntax edges NextToken and Child (ablation) | Program Graph · SeenProjTest VarMisuse | accuracy (%): 78.4 -> 85.5 | Helped |
| add syntax edges NextToken and Child (ablation) | Program Graph · SeenProjTest VarNaming | accuracy (%): 52.9 -> 53.6 | No clear effect |
| add semantic edges (ablation) | Program Graph · SeenProjTest VarMisuse | accuracy (%): 55.3 -> 85.5 | Helped |
| add semantic edges (ablation) | Program Graph · SeenProjTest VarNaming | accuracy (%): 34.3 -> 53.6 | Helped |
| use subtokens instead of whole tokens for node labels (ablation) | Program Graph · SeenProjTest VarNaming | accuracy (%): 34.5 -> 53.6 | Helped |
| use subtokens instead of whole tokens for node labels (ablation) | Program Graph · SeenProjTest VarMisuse | accuracy (%): 85.6 -> 85.5 | No clear effect |
Closest papers by meaning
- GraphCodeBERT: Pre-training Code Representations with Data Flow (ICLR 2021 · similarity 0.89)
- Open Vocabulary Learning on Source Code with a Graph-Structured Cache (ICML 2019 · similarity 0.88)
- Multi-View Graph Representation for Programming Language Processing: An Investigation into Algorithm Detection (AAAI 2022 · similarity 0.87)
- Emergent Representations of Program Semantics in Language Models Trained on Programs (ICML 2024 · similarity 0.85)
- VQGraph: Rethinking Graph Representation Space for Bridging GNNs and MLPs (ICLR 2024 · similarity 0.85)
- Chain of Execution Supervision Promotes General Reasoning in Large Language Models (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.