Research map / Graph learning / Knowledge and temporal graphs
TFLEX: Temporal Feature-Logic Embedding Framework for Complex Reasoning over Temporal Knowledge Graph
Xueyuan Lin, Haihong E, Chengjin Xu, Gengxian Zhou, Haoran Luo, Tianyi Hu, Fenglong Su, Ningyuan Li, Mingzhi Sun
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 |
|---|---|---|---|
| replace fuzzy entity part with ConE geometric entity part (X(ConE)) (ablation) | TFLEX · ICEWS14 | avg_e MRR: 56.79 -> 40.93 | Hurt |
| remove the time part (FLEX) (ablation) | TFLEX · ICEWS14 | avg_e MRR: 56.79 -> 43.67 | Hurt |
| remove the time part (FLEX) (ablation) | TFLEX · GDELT-500 | avg_e MRR: 19.60 -> 19.07 | Hurt |
| merge entity and timestamp features into one feature (X-1F) (ablation) | TFLEX · ICEWS05-15 | avg_e MRR: 48.99 -> 49.90 | Improved |
| merge entity and timestamp features into one feature (X-1F) (ablation) | TFLEX · GDELT-500 | avg_e MRR: 19.60 -> 17.92 | Hurt |
| merge entity and timestamp features into one feature (X-1F) (ablation) | TFLEX · ICEWS14 | avg_e MRR: 56.79 -> 56.89 | No clear effect |
| remove entity and time logic (X-logic) (ablation) | TFLEX · ICEWS05-15 | avg_e MRR: 48.99 -> 44.80 | Hurt |
| remove entity and time logic (X-logic) (ablation) | TFLEX · GDELT-500 | avg_e MRR: 19.60 -> 17.36 | Hurt |
| remove entity and time logic (X-logic) (ablation) | TFLEX · ICEWS14 | avg_e MRR: 56.79 -> 56.64 | No clear effect |
| train on temporal complex queries instead of only one-hop queries (TFLEX vs TFLEX-1p) (ablation) | TFLEX · TKGC (ICEWS14, ICEWS05-15, GDELT) | MRR: 43.9 -> 48.2 | Improved |
| use TFLEX instead of ConE baseline for query answering | ConE · ICEWS14 | avg_e MRR: 41.94 -> 56.79 | Improved |
| embedding dimension d and margin gamma sweep | TFLEX · ICEWS14, ICEWS05-15, GDELT-500 | MRR | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ConE — ConE: Cone Embeddings for Multi-Hop Reasoning over Knowledge Graphs (NeurIPS 2021) · better on 3 of 3 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
- Rethinking Complex Queries on Knowledge Graphs with Neural Link Predictors (ICLR 2024 · similarity 0.89)
- Fuzzy Logic Based Logical Query Answering on Knowledge Graphs (AAAI 2022 · similarity 0.88)
- Neural Methods for Logical Reasoning over Knowledge Graphs (ICLR 2022 · similarity 0.87)
- NQE: N-ary Query Embedding for Complex Query Answering over Hyper-Relational Knowledge Graphs (AAAI 2023 · similarity 0.87)
- Temporal Conjunctive Query Answering via Rewriting (AAAI 2025 · similarity 0.85)
- TEILP: Time Prediction over Knowledge Graphs via Logical Reasoning (AAAI 2024 · 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.
