Research map / Graph learning / Knowledge and temporal graphs
Inductive knowledge graph completion
Complete knowledge graphs for unseen entities using relational semantics and GNNs.
38 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Knowledge and temporal graphs, keyword cluster “knowledge graph · entities · kg”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Leveling off Several papers from 2025–26 still report on 2 tracked benchmarks, but the best result has barely moved since 2024.
10 papers in 2023–24 and 12 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 3.9% to 4.2%.
What looks settled
- WN18RR (MRR): best up to 2024 0.781, best in 2025–26 0.501 from 2 papers, at most 0.2% better, if at all.
- FB15k-237 (MRR): best up to 2024 0.522, best in 2025–26 0.428 from 2 papers, at most 0.2% better, if at all.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2018 · Open-World Knowledge Graph Completion (AAAI 2018 · 284 citations)
- 2020 · Diachronic Embedding for Temporal Knowledge Graph Completion (AAAI 2020 · 386 citations)
- 2021 · Type-augmented Relation Prediction in Knowledge Graphs (AAAI 2021 · 48 citations)
- 2022 · Temporal Knowledge Graph Completion Using Box Embeddings (AAAI 2022 · 98 citations)
- 2023 · ExpressivE: A Spatio-Functional Embedding For Knowledge Graph Completion (ICLR 2023 · 12 citations)
- 2024 · LAFA: Multimodal Knowledge Graph Completion with Link Aware Fusion and Aggregation (AAAI 2024 · 30 citations)
- 2025 · APKGC: Noise-enhanced Multi-Modal Knowledge Graph Completion with Attention Penalty (AAAI 2025 · 12 citations)
- 2026 · Information-Theoretic Minimal Sufficient Representation for Multi-Domain Knowledge Graph Completion (AAAI 2026 · 2 citations)
Earliest papers here
- Open-World Knowledge Graph Completion (AAAI 2018 · 284 citations)
- Neural Knowledge Acquisition via Mutual Attention Between Knowledge Graph and Text (AAAI 2018 · 164 citations)
- Data-Dependent Learning of Symmetric/Antisymmetric Relations for Knowledge Base Completion (AAAI 2018 · 8 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
WN18RR · MRR (higher is better)
10 papers, 2020–2025 · flat since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 0.498 | Duality-Induced Regularizer for Tensor Factorization Based Knowledge Graph Completion (NeurIPS 2020 · 26 citations) |
| 2022 | 0.499 | ER: Equivariance Regularizer for Knowledge Graph Completion (AAAI 2022 · 10 citations) |
| 2023 | 0.508 | ExpressivE: A Spatio-Functional Embedding For Knowledge Graph Completion (ICLR 2023 · 12 citations) |
| 2024 | 0.781 | Knowledge Graphs Can be Learned with Just Intersection Features (ICML 2024) |
FB15k-237 · MRR (higher is better)
9 papers, 2020–2025 · flat since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 0.522 | Relational Graph Neural Network with Hierarchical Attention for Knowledge Graph Completion (AAAI 2020 · 190 citations) |
YAGO3-10 · MRR (higher is better)
6 papers, 2020–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 0.584 | Duality-Induced Regularizer for Tensor Factorization Based Knowledge Graph Completion (NeurIPS 2020 · 26 citations) |
| 2022 | 0.588 | ER: Equivariance Regularizer for Knowledge Graph Completion (AAAI 2022 · 10 citations) |
| 2024 | 0.617 | Knowledge Graphs Can be Learned with Just Intersection Features (ICML 2024) |
What has been tried
32 papers here report 326 measured changes (a component added, removed or swapped, compared with a baseline): helped 71%, no clear effect 5%, hurt 22%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add modality interaction attention instead of traditional vector similarity matching | FB15k-237-IMG · Hits@10 | 0.538 -> 0.551 | Helped | LAFA: Multimodal Knowledge Graph Completion with Link Aware Fusion and Aggregation (AAAI 2024 · 30 citations) |
| add the mixed geometry message function | FB15k-237, YAGO3-10, WN18RR · MRR | .377 -> .393; .568 -> .586; .496 -> .511 | Helped | Mixed Geometry Message and Trainable Convolutional Attention Network for Knowledge Graph Completion (AAAI 2024 · 19 citations) |
| use all modalities instead of structure only | DB15K · Hits@1 | 0.125 -> 0.282 | Helped | APKGC: Noise-enhanced Multi-Modal Knowledge Graph Completion with Attention Penalty (AAAI 2025 · 12 citations) |
| Use APs with detailed descriptions instead of CPs only with detailed descriptions | WN18RR, FB15k-237, and NELL-995, transductive and inductive · MRR | Helped | Anchoring Path for Inductive Relation Prediction in Knowledge Graphs (AAAI 2024 · 10 citations) | |
| combine anchor-enhanced and relation-aware embeddings rather than use relation-aware embedding alone | WN18RR · MRR | 55.31 -> 59.74 | Helped | Knowledge Graph Completion with Relation-Aware Anchor Enhancement (AAAI 2025 · 8 citations) |
| replace connection-biased attention with vanilla attention | WN18RR v1 · Hits@10 | 97.30 -> 92.00 | Hurt | Replacing Paths with Connection-Biased Attention for Knowledge Graph Completion (AAAI 2025 · 4 citations) |
| add Semantic Smoothing | WN18RR (Avg.) · Hits@10 | 78.49 -> 81.23 | Helped | S²DN: Learning to Denoise Unconvincing Knowledge for Inductive Knowledge Graph Completion (AAAI 2025 · 3 citations) |
| Add the transformation function T (log function) to CRR | FB15k-237 · MRR | 0.414 -> 0.428 | Helped | KGCRR: An Effective Metric-Driven Knowledge Graph Completion Framework by Designing a Novel Upper Bound Function with Adaptive Approximation to Reciprocal Rank (AAAI 2025 · 1 citation) |
| Add the intermediate-output gating mechanism | FB15k-237 · Hits@1 | Helped | SPAC: Sparse Partitioning and Adaptive Core Tensor Pruning Model for Knowledge Graph Completion (AAAI 2025 · 1 citation) | |
| Increase RTMD depth from 2 to 3 layers | WNv3 · classification accuracy | 0.8204 -> 0.8205 | No clear effect | Stability and Generalization Capability of Subgraph Reasoning Models for Inductive Knowledge Graph Completion (ICML 2025 · 0 citations) |
| Use only IVR upper bound 1 instead of full IVR | WN18RR, FB15k-237, YAGO3-10 · MRR | 0.501 -> 0.497; 0.368 -> 0.366; 0.581 -> 0.576 | Hurt | Knowledge Graph Completion by Intermediate Variables Regularization (NeurIPS 2024 · 0 citations) |
| replace relation-weighted aggregation with mean aggregation | FB-100 and NL-100 · MRR | 0.244 vs 0.253; 0.279 vs 0.314 | Hurt | Towards Global-Topology Relation Graph for Inductive Knowledge Graph Completion (AAAI 2025) |
What reviewers will expect
Datasets and baselines named in the 9 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- FB15k-237 · 6 of 9 papers
- WN18RR · 5 of 9 papers
- NELL995 · 3 of 9 papers
Baselines
- GraIL · 4 of 9 papers
- TransE · 3 of 9 papers
- DistMult · 3 of 9 papers
- RotatE · 3 of 9 papers
- NeuralLP · 3 of 9 papers
- DRUM · 3 of 9 papers
- CoMPILE · 3 of 9 papers
- NBFNet · 3 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 22 papers; from OpenAlex and the papers' affiliations).
- Xi’an Jiaotong University · 2 papers
- Xidian University · 2 papers
- Korea Advanced Institute of Science and Technology · 2 papers
Use it
Download BibTeX of 34 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Knowledge graph embeddings (110 papers)
- Temporal knowledge graph reasoning (84 papers)
- Logical query answering and rule reasoning (68 papers)
Elsewhere in the topic
- Continuous-time dynamic graph learning (132 papers)
- Graph reasoning for visual recognition (96 papers)
- Graph-based dialog systems (94 papers)
- Social and behavior graphs (87 papers)
- Scene graph generation (70 papers)
- Temporal graph neural architectures (69 papers)
- Heterogeneous graph learning (66 papers)
- Spatio-temporal graph traffic forecasting (60 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).