A Survey of Knowledge Graph Reasoning on Graph Types: Static, Dynamic, and Multimodal

TL;DR

<3-5 sentence high-level summary> This survey addresses knowledge graph reasoning across static, temporal, and multi-modal knowledge graphs by adopting a bi-level taxonomy that separates graph types from techniques and reasoning scenarios. It comprehensively reviews embedding-, path-, and rule-based methods for static KGR; RNN-based and RNN-agnostic approaches for temporal KGR; and transformer-based versus transformer-agnostic strategies for multi-modal KGR, while also detailing datasets, challenges, and practical applications. The work highlights trends toward inductive and scalable reasoning, the need for explainability, and the potential for integration with large language models to enhance reasoning and adoption. An accompanying open-source GitHub repository aggregates 180 models and 67 datasets to support reproducibility and community building.

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. The corresponding open-source repository is shared on GitHub https://github.com/LIANGKE23/Awesome-Knowledge-Graph-Reasoning.

Figures (13)

• Figure 1: Examples of three categories of the knowledge graphs, i.e., static, temporal, and multi-modal knowledge graph.
• Figure 2: Overview framework of the survey.
• Figure 3: Comparison between two types of multi-modal knowledge graphs. N-MMKG represents the multi-modal data as entities, while A-MMKG represents multi-modal data as new attributes.
• Figure 4: Illustration of transductive and inductive reasoning. In the transductive scenario, entities in test graphs are all seen during the training procedure. While as for the inductive scenario, unseen entities may exist in test graphs.
• Figure 5: Illustration of interpolation and extrapolation reasoning. The timestamp $t$ for knowledge graph reasoning in the interpolation scenario is seen in the past ($0 \leq t \leq T$). While the queried facts in the future ($t \geq T$) for the extrapolation scenario.

Theorems & Definitions (3)

• Definition 1
• Definition 2
• Definition 3