Research on an Advertising Click-Through Rate Prediction Algithm Combining Attention Mechanism Time-Gated Recurrent Unit

Click-through rate (CTR) prediction critical task in online advertising, influencing ad placement strategies revenue generation. Machine learning models struggle with feature sparsity sequential dependencies in user behave data limiting predictive perfmance. Deep learning methods recurrent neural netwks (RNNs) improved CTR prediction capturing sequential patterns, often fail to adequately model long-range dependencies time-sensitive interactions. Address challenges that proposes CTR prediction algithm that integrates an attention mechanism with a Time-Gated Recurrent Unit (TGRU) enhances the conventional GRU incpating tempal gating, the model to adaptively capture time-dependent variations user interactions. Experimental evaluations public CTR datasets demonstrate that the proposed model outperfms state-of-the-art approaches higher prediction accuracy better generalization. Further analysis ablation studies attention weight visualization confirms the effectiveness our method in capturing key temp, behavial patternsf me accurate CTR estimation provides a promising direction optimizing CTR prediction in dynamic advertising environments.