<p>Temporal knowledge graph (TKG) reasoning, which involves predicting future facts by leveraging existing data within a TKG, has garnered significant attention in recent research. While existing methods primarily rely on embedding models to address entity prediction, they are generally unable to handle relation prediction concurrently. Moreover, these approaches struggle to effectively differentiate the significance of entities and relations within the topological structure, which hinders their ability to capture the inherent semantics in knowledge graph. To overcome these limitations, this paper introduces a novel TKG reasoning model called NHIA, which integrates neighboring and historical information aggregation to handle both entity and relation prediction. NHIA mainly comprises two key modules: the neighbor aggregation module and the history aggregation module. The former leverages the graph neural network to aggregate relevant entities and relations to generate target entity representations for each snapshot. The latter introduces two gates to learn evolve-aware entity embeddings, considering both current and previous embeddings. Meanwhile, the gate recurrent unit is adopted to capture evolve-aware relation embeddings. Experimental results across three benchmark datasets validate the effectiveness of NHIA, demonstrating its superior performance in both entity and relation prediction compared to state-of-the-art baselines. Notably, NHIA achieves up to a 0.61% improvement in entity prediction and a 2.53% improvement in relation prediction, underscoring its capacity to capture the temporal and relational complexities inherent in TKGs.</p>