Recent advances in Knowledge Graph Completion

Knowledge Graph Completion is dedicated to filling the gaps in knowledge graphs, and this technology has found widespread application in various domains. Traditional knowledge graph completion techniques operate under a closed-world assumption, restricting their analysis to existing content within the same knowledge graph, making it impossible to introduce new entities and relationships. Furthermore, most methods are designed to handle a single type of information, primarily focusing on temporal knowledge graph completion. However, in the real open world, knowledge graphs are dynamic, encompassing multimodal data that evolves over time. As such, traditional knowledge graph completion methods may struggle to address practical needs. To address these limitations, this paper provides an overview of traditional knowledge graph completion research. It highlights the latest advancements in knowledge graph completion methods based on open-world, multimodal, and temporal knowledge. We analyze the enhancements brought about by these new methods in the field of knowledge graph completion, offering a comparative analysis that explores their strengths, weaknesses, application domains, and future development trends to serve as a valuable reference for researchers.