Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 357 of the 357 references it cites were found in a registry or an index; none of the 78 sources its claims cite is checked for support yet.

A Survey of Knowledge Graph Reasoning on Graph Types: Static, Dynamic, and Multi-Modal

Open access
Liang, Ke, Meng, Lingyuan, Liu, Meng, Liu, Yue, Tu, Wenxuan, Wang, Siwei, Zhou, Sihang, Liu, Xinwang, Sun, Fuchun, He, Kunlun · 2024 · IEEE Transactions on Pattern Analysis and Machine Intelligence

Abstract

Knowledge graph reasoning (KGR), aiming to deduce new facts from existing facts based on mined logic rules underlying knowledge graphs (KGs), has become a fast-growing research direction. It has been proven to significantly benefit the usage of KGs in many AI applications, such as question answering, recommendation systems, and etc. According to the graph types, existing KGR models can be roughly divided into three categories, i.e., static models, temporal models, and multi-modal models. Early works in this domain mainly focus on static KGR, and recent works try to leverage the temporal and multi-modal information, which are more practical and closer to real-world. However, no survey papers and open-source repositories comprehensively summarize and discuss models in this important direction. To fill the gap, we conduct a first survey for knowledge graph reasoning tracing from static to temporal and then to multi-modal KGs. Concretely, the models are reviewed based on bi-level taxonomy, i.e., top-level (graph types) and base-level (techniques and scenarios). Besides, the performances, as well as datasets, are summarized and presented. Moreover, we point out the challenges and potential opportunities to enlighten the readers.

Overview

CoChat Summary

TL;DR

This survey reviews knowledge graph reasoning (KGR) across static, temporal, and multi-modal KGs using a bi-level taxonomy of graph types and base-level techniques/scenarios. It summarizes 90 static, 58 temporal, and 32 multi-modal KGR models, plus datasets and performance comparisons, and highlights open challenges.

Key findings

  • The survey organizes KGR into three graph types, fourteen techniques, and four reasoning scenarios; it reviews 90 static KGR models, 58 temporal KGR models, and 32 multi-modal KGR models.
  • Static KGR: embedding-based models generally have better expressive ability but lack explainability; GNN-based models are considered most promising, while path-based models suit transductive reasoning and rule-based models naturally support inductive reasoning. Of the static models reviewed, there are 56 transductive and 34 inductive models.
  • Temporal KGR: RNN-based models are considered most promising and are more easily adopted to extrapolation; extrapolation reasoning is still early, occupying only around 30% of temporal KGR models. Reported results include TeMP achieving MRR 69.10 and H@10 91.70 on ICEWS05-15, and RPC achieving MRR 22.41 on GDELT.
  • Multi-modal KGR: research is still early, with only 18% of models for multi-modal scenarios; reported results include IMF at H@1 28.7 and H@10 59.3 on FB15k-237-IMG, and MoSE at H@1 94.8 and H@10 97.4 on WN18-IMG. The survey also shares 180 state-of-the-art KGR models and 67 typical datasets on GitHub.

Methods

  • Literature survey design: the paper defines static, temporal, and multi-modal KGs, formulates KGR tasks, and introduces transductive/inductive and interpolation/extrapolation scenarios.
  • Models are reviewed under a bi-level taxonomy: top-level graph types (static, temporal, multi-modal) and base-level techniques and scenarios.
  • The survey summarizes typical datasets and performance tables for static, temporal, and multi-modal KGR, and points out challenges and opportunities; an open-source repository accompanies the collection.

Limitations

  • The paper does not explicitly state limitations of the survey itself.
  • Evident from the text: as a survey, it conducts no new experiments or unified benchmark evaluation; performance comparisons are compiled from prior reported results, and some table entries are missing.
  • The text itself notes that several areas remain early-stage, including extrapolation reasoning, multi-modal KGR, out-of-distribution reasoning, and large-scale reasoning.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/tpami.2024.3417451

Cites

357 of 357 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 Valid · 7 Not valid · 61 Not checked · 85 Own results and interpretations

36 more claims

Create a free account to see all 153 claims and to verify the 60 references still open.

See all claims — free account
  • It may state two separate claims as one. 45
  • Probably not a substantial claim. 86
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 336 references

Research works printed but not linked: 12

  1. Ren, Query2Box: Reasoning over knowledge graphs in vector space using box embeddings, 1
  2. 10.5555/3524938.3525087 · doi:10.5555/3524938.3525087
  3. Huang, A theory of link prediction via relational weisfeiler-leman
  4. Cheng, Differentiable neuro-symbolic reasoning on large-scale knowledge graphs
  5. Zhang, Temporal knowledge graph completion with approximated Gaussian process embedding, 4697
  6. Sadeghian, Temporal reasoning over event knowledge graphs, 1
  7. 10.18653/v1/d18-1516 · DOI · Paper page
  8. 10.18653/v1/2020.emnlp-main.462 · DOI · Paper page
  9. 10.18653/v1/d19-1269 · DOI
  10. 10.18653/v1/d17-1060 · DOI · Paper page
  11. Mahdisoltani, 2013, YAGO3: A knowledge base from multilingual Wikipedias
  12. 10.18653/v1/d18-1516 · DOI · Paper page

Cited by

Items 301–400 of 553

Metadata

Authors
Liang, K., Meng, L., Liu, M., Liu, Y., Tu, W., Wang, S., Zhou, S., Liu, X., Sun, F., & He, K.
Affiliations
  • National University of Defense Technology (CN)
  • Tsinghua University (CN)
  • Chinese PLA General Hospital (CN)
Venue
IEEE Transactions on Pattern Analysis and Machine Intelligence
Published
2024-12
Volume
46
Issue
12
Pages
9456-9478
ISSN
  • 0162-8828 (ISSN-L)
  • 2160-9292
  • 1939-3539
OpenAlex
W4400111732
Cited by
553
References
336
Citations by year
2026: 208 · 2025: 188 · 2024: 114
FWCI
42.52
Type
journal article
Language
en
Open access
Locations
  • IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Funders
National Natural Science Foundation of China (62325604, 62276271)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Advanced Graph Neural Networks
  • Physical Sciences › Computer Science › Artificial Intelligence › Semantic Web and Ontologies
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
Keywords
knowledge graph reasoning, temporal knowledge graphs, multimodal knowledge graphs, temporal reasoning, multimodal reasoning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-18, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-18 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.