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

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.

I. Introduction

Humans learn skills from two main sources, i.e., specialized books and working experiences. For example, a good doctor needs to get knowledge from school and practice experiences from the hospital. However, most existing artificial intelligence (AI) models only imitate the learning procedure from experiences while ignoring the former [1], [2], thus making them less explainable and worse performances. Knowledge graphs (KGs), which store the human knowledge facts in intuitive graph structures [3], are treated as potential solutions these years. While, the construction of KGs is a dynamic and continuous procedure, thus most KGs suffer from incomplete issues, hindering their effectiveness in KG-assisted applications, such as question answering [4], recommendation system [5]. To alleviate the problem, knowledge graph reasoning (KGR) has drawn increasing attention these years. It aims to infer missing facts from existing ones in KGs. Taking Fig. 1(a) as the target KG, KGR models are expected to derive the logic rules (A, father of, B)$\wedge {(A, husband\ of, C)}\rightarrow$(C, mother of, B), and then further infer the missing fact (Savannah, mother of, Bronny).