Historical Trends and Normalizing Flow for One-shot Temporal Knowledge Graph Reasoning

Ruixin Ma, Longfei Wang, Huinan Wu, Buyun Gao, Xiaoru Wang, Liang Zhao

Abstract: Highlights•Utilizing a domain graph aggregator, initial entity representations are obtained.•The out-of-distribution problem is mitigated through the normalizing flow technique.•Trend information serves as the primary modeling object.•The similarity of facts is analyzed through two distinct perspectives.