Research map / Graph learning / Knowledge and temporal graphs
Expanding Holographic Embeddings for Knowledge Completion
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 |
|---|---|---|---|
| expand HolE with 16 random 0/1 perturbation vectors (ablation) | HolE dim=256 · FB15K | HITS@10 (%): 83.0 -> 88.6 | Helped |
| increase random 0/1 vectors from 2 to 8 (ablation) | HolEx · FB15K | HITS@10 (%): 85.4 -> 87.9 | Helped |
| use 8 Haar vectors instead of 8 random 0/1 vectors (ablation) | HolEx · FB15K | HITS@10 (%): 87.9 -> 86.7 | Hurt |
| expand HolE with 8 Haar vectors (ablation) | HolE dim=256 · FB15K | HITS@10 (%): 83.0 -> 86.7 | Helped |
| increase embedding dimension from 150 to 256 (ablation) | HolE reimplementation · FB15K | HITS@10 (%): 78.4 -> 83.0 | Helped |
| increase perturbation vectors l from 2 to 8 (ablation) | HolEx · FB15K | Mean Rank: 51 -> 48 -> 47 | No clear effect |
| use identity-matrix truncation as perturbation vectors (ablation) | HolEx · FB15K | HITS@10 (%) | Hurt |
| reimplement HolE in ProjE framework (entity-relation interaction, dropout) (ablation) | HolE dim=150 · FB15K | HITS@10 (%): 73.9 -> 78.4 | Helped |
| increase perturbation vectors (training cost) (ablation) | HolEx · FB15K | time per epoch | Hurt |
| HolEx 16 random vs re-tuned DistMult | HolEx · FB15K | HITS@10 (%): 89.3 vs 88.6 | Hurt |
Closest papers by meaning
- Holographic Embeddings of Knowledge Graphs (AAAI 2016 · similarity 0.92)
- Improved Knowledge Graph Embedding Using Background Taxonomic Information (AAAI 2019 · similarity 0.88)
- LENA: Locality-Expanded Neural Embedding for Knowledge Base Completion (AAAI 2019 · similarity 0.88)
- ExpressivE: A Spatio-Functional Embedding For Knowledge Graph Completion (ICLR 2023 · similarity 0.88)
- SPAC: Sparse Partitioning and Adaptive Core Tensor Pruning Model for Knowledge Graph Completion (AAAI 2025 · similarity 0.86)
- Replacing Paths with Connection-Biased Attention for Knowledge Graph Completion (AAAI 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.