时序知识图谱构建关键技术及研究进展

Key technologies and research progress of temporal knowledge graph construction

中文关键词  时序知识图谱  时序知识抽取  时序知识嵌入  时序逻辑推理  多粒度  深度学习

英文关键词  temporal knowledge graph  temporal knowledge extraction  temporal information embedding  temporal logical reasoning  multi-granularity  deep learning

中文摘要

      近年来, 时序知识图谱受到国内外研究人员的广泛关注, 成为研究热点. 区别于传统知识图谱, 时序知识图 谱嵌入了时序信息, 能够表示时序关系和因果关系, 描述关系变化规律和事件演化模式, 可成为能感知时间信息和 因果律的知识库. 本文对时序知识图谱构建技术进行综述. 首先, 本文介绍了时序知识图谱的定义和构建流程; 其 次, 梳理了时序知识图谱构建过程中时序知识抽取、时序知识嵌入、时序逻辑推理等关键技术的研究进展, 分析其 发展状况并指出优缺点; 最后, 给出面向多级决策的铝电解氧化铝浓度多粒度知识图谱构建与应用案例, 对时序知 识图谱在流程工业领域的应用前景进行了展望.

英文摘要

      In recent years, temporal knowledge graphs have garnered significant attention from both academia and industry, emerging as a major research hotspot. Unlike the traditional knowledge graph, the temporal knowledge graph embed temporal information, enabling representation of temporal relationship and causal relationships, describe the laws of relationship changes and event evolution patterns, and can serve as a knowledge base that can perceiving temporal information and causal laws. This paper primarily reviews the construction technologies of temporal knowledge graph. Firstly, the article introduces the definition and construction process of temporal knowledge graph. Secondly, it sorts out the research progress of key technologies such as temporal knowledge extraction, temporal knowledge embedding, and temporal logic reasoning in the process of temporal knowledge graph construction, analyzes its development status and points out its advantages and disadvantages. Finally, presents the multilevel decision of aluminum electrolysis alumina concentration multi-granularity knowledge graph construction and application case, the application prospects of temporal knowledge graph in process industry are prospected.