<![CDATA[The rapid growth of electronic services has created significant opportunities for personalized product recommendations through artificial intelligence (AI) systems. However, existing recommendation algorithms face critical challenges, including scalability, cold-start issues, and performance degradation in big data environments. This research performs a systematic review of 73 studies published from 2022 until 2024 to examine AI architecture frameworks applied to product analysis and recommendation systems in electronic service. The review identifies dominant frameworks such as CNN, RNN/LSTM, TensorFlow, Spark, and emerging technologies like GNN, alongside distributed infrastructures such as Hadoop for large-scale data processing. Research methods observed include experiments, benchmarks, simulations, surveys, and case studies. Key findings emphasize performance and efficiency improvements, accuracy, and scalability concerns. Based on these insights, this paper proposes a multi-layered AI architecture framework integrating data ingestion, distributed storage, model development, MLOps orchestration, privacy-preserving learning, and adaptive feedback loops. The proposed framework addresses scalability and sustainability challenges while ensuring high-performance recommendation capabilities. This study contributes a comprehensive blueprint for organizations seeking to deploy robust, scalable, and privacy-aware AI systems in dynamic e-service environments.]]>
Copyrights © 2026
