Research map / Graph learning / Knowledge and temporal graphs
Interstellar: Searching Recurrent Architecture for Knowledge Graph Embedding
Yongqi Zhang, Quanming Yao, Lei Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Search a task-specific recurrent architecture over relational paths | KG entity alignment · DBP-WD Normal, DBP-YG Normal, EN-FR Normal, EN-DE Normal | MRR: DBP-WD: 0.49 -> 0.51; DBP-YG: 0.50 -> 0.51; EN-FR: 0.44 -> 0.46; EN-DE: 0.57 -> 0.59 | Improved |
| Search a task-specific recurrent architecture over relational paths | KG link prediction · WN18-RR and YAGO3-10 | MRR: WN18-RR: 0.47 -> 0.48; YAGO3-10: 0.48 -> 0.51 | Improved |
| Search a task-specific recurrent architecture over relational paths | KG link prediction · FB15k-237 | MRR: 0.32 -> 0.32 | No clear effect |
| Use longer relational-path modeling with intermediate entities and relations (ablation) | Countries S3 · Countries S3 | AUC-PR: P1 0.933±0.031; P2 0.952±0.023; P3 0.961±0.016; P4 0.964±0.015; Interstellar 0.968± 0.007 | Improved |
| Include entities along the path rather than only relations (ablation) | Countries relational-path models · Countries S1 and S2 | AUC-PR: S1: P3 0.992±0.001 -> P4 0.977±0.028; S2: P3 1.000±0.000 -> P4 0.984±0.010 | Hurt |
| Use a one-shot parameter-sharing search across the full architecture space (ablation) | Interstellar architecture search · Entity alignment and link prediction search | validation performance | Hurt |
| Split macro and micro architecture search, evaluating macro candidates stand-alone and micro candidates with parameter sharing | Interstellar architecture search · Entity alignment and link prediction search | search efficiency and reliability | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- RotatE — RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space (ICLR 2019) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning to Exploit Long-term Relational Dependencies in Knowledge Graphs (ICML 2019 · similarity 0.84)
- You CAN Teach an Old Dog New Tricks! On Training Knowledge Graph Embeddings (ICLR 2020 · similarity 0.84)
- LENA: Locality-Expanded Neural Embedding for Knowledge Base Completion (AAAI 2019 · similarity 0.84)
- einspace: Searching for Neural Architectures from Fundamental Operations (NeurIPS 2024 · similarity 0.84)
- Towards Global-Topology Relation Graph for Inductive Knowledge Graph Completion (AAAI 2025 · similarity 0.84)
- Integrating Large Language Models and Möbius Group Transformations for Temporal Knowledge Graph Embedding on the Riemann Sphere (AAAI 2025 · similarity 0.83)
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.
