Research map / Graph learning / Graph self-supervised learning
Dynamic Embedding on Textual Networks via a Gaussian Process
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 |
|---|---|---|---|
| combine textual and structural embeddings rather than using text alone (ablation) | DetGP (Wavg) · Cora link prediction, 15% training edges | AUC: 83.4 -> 92.8 | Helped |
| combine textual and structural embeddings rather than using structure alone (ablation) | DetGP (Wavg) · Cora link prediction, 15% training edges | AUC: 85.4 -> 92.8 | Helped |
| use textual embeddings rather than structural embeddings alone (ablation) | DetGP (Wavg) · Cora node classification, 10% training nodes | Macro-F1: 70.9 -> 78.1 | Helped |
| use DWavg rather than Wavg as the text encoder (ablation) | DetGP · Cora node classification, 10% training nodes | Macro-F1: 80.5 -> 83.1 | Helped |
| use mean neighbor aggregation rather than only text | dynamic embedding baseline · Cora dynamic link prediction, 10% training nodes | AUC: 61.2 -> 61.8 | Helped |
| use max-pooling neighbor aggregation rather than only text | Neighbor-Aggregate · Cora dynamic link prediction, 10% training nodes | AUC: 61.2 -> 54.6 | Hurt |
| use GraphSAGE mean aggregation rather than only text | dynamic embedding baseline · Cora dynamic link prediction, 10% training nodes | AUC: 61.2 -> 62.2 | Helped |
| infer new-node structure with the GP and learned inducing points rather than aggregate neighbors | DetGP · Cora dynamic link prediction, 10% training nodes | AUC: 62.2 -> 62.9 | Helped |
| infer new-node structure with the GP and learned inducing points rather than aggregate neighbors | DetGP · DBLP dynamic node classification, 10% training nodes | Macro-F1: 59.0 -> 60.2 | Helped |
| use graph structure alongside text for classification (ablation) | DetGP (Wavg) · Cora node classification, 10% training nodes | Macro-F1: 78.1 -> 80.5 | Helped |
Closest papers by meaning
- Unifying Text Semantics and Graph Structures for Temporal Text-attributed Graphs with Large Language Models (NeurIPS 2025 · similarity 0.88)
- DTGB: A Comprehensive Benchmark for Dynamic Text-Attributed Graphs (NeurIPS 2024 · similarity 0.87)
- TGLsta: Low-resource Textual Graph Learning with Semantic and Topological Awareness via LLMs (AAAI 2025 · similarity 0.87)
- Towards Dynamic Message Passing on Graphs (NeurIPS 2024 · similarity 0.86)
- Generalization Principles for Inference over Text-Attributed Graphs with Large Language Models (ICML 2025 · similarity 0.86)
- GNNs Meet Sequence Models Along the Shortest-Path: an Expressive Method for Link Prediction (NeurIPS 2026 · similarity 0.86)
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.