Analysis and Optimization of Ranking Patterns in Advertising Platforms Using Explainable Artificial Intelligence (XAI) and SEO Optimization
Volume 2, Issue 2, March and April 2026, Pages 127-138
https://doi.org/10.5281/zenodo.20847500
Ali Salehi Vojdeh Nazari, Mahdi Baradaran
Abstract The aim of this study is to design and evaluate a hybrid model based on SEO metrics and explainable machine learning algorithms (XAI) to improve the ranking of digital advertisements. The research data consisted of 5,000 simulated records, which after preprocessing were analyzed using XGBoost, Random Forest, and Elastic Net models. Performance evaluation using metrics such as Accuracy, MAE, and RMSE indicated that the XGBoost model outperformed the others. To understand the model's decision-making logic, SHAP and LIME were employed. The results highlight the significant role of content quality, keyword relevance, and click-through rate in determining ad ranking. Additionally, fidelity and stability metrics of explanations showed that the best-performing model not only offers appropriate accuracy but also exhibits high transparency and stability in providing explanations. In the structural analysis, the mediating role of "user trust" in strengthening the effect of SEO metrics was confirmed. Overall, the findings suggest that combining SEO metrics with explainable models can lead to both improved prediction accuracy and enhanced transparency and trustworthiness in ad ranking.