ChronoBridge: A Novel Framework for Enhanced Temporal and Relational Reasoning in Temporal Knowledge Graphs

The prediction task of entities and relationships in Temporal Knowledge Graph (TKG) extrapolation is crucial and extensively studied. Mainstream algorithms like Gated Recurrent Unit (GRU) models primarily focus on encoding historical factual features within TKGs, often neglecting the importance of incorporating entity and relationship features during decoding. This bias ultimately leads to loss of detail and inadequate prediction accuracy during the inference process .Addressing this issue, a novel ChronoBridge framework is proposed, featuring a dual mechanism of a Chronological node Encoder and a Bridged Feature Fusion Decoder. Specifically, the Temporal Node Encoder employs an advanced recursive neural network with enhanced GRU in an autoregressive manner to model historical KG sequences, thereby accurately capturing entity changes over time and significantly enhancing the model’s ability to identify and encode temporal patterns of facts across the timeline. Meanwhile, the Bridge Feature Fusion Decoder utilizes a new variant of GRU and a multi-layer perceptron mechanism during the prediction phase to extract entity and relationship features and fuse them for inference, thereby strengthening the model’s reasoning capabilities for future events.Test results on three standard datasets demonstrate significant performance improvements compared to existing techniques, with a 25.21% increase in Mean Reciprocal Rank (MRR) accuracy for prediction tasks and a 39.38% improvement in relationship inference, thus validating its effectiveness. This breakthrough not only enhances understanding of temporal evolution in knowledge graphs but also paves the way for future research and applications in TKG reasoning.