The Research Department of Economics and Management of Tadbir Nikan Business, Marketing, and Finance Open 3092-6238 1 Serial Number 3 2024 06 10 Designing an Intelligent Framework for Real-Time Recommendation Based on Momentary User Behavior Analysis in Online Platforms Designing an Intelligent Framework for Real-Time Recommendation Based on Momentary User Behavior Analysis in Online Platforms 159 168 10.61838/bmfopen.1.3.14 EN Journal Article 2024 03 07 <p>The objective of this study was to design and empirically evaluate an intelligent real-time recommender framework that leverages momentary user behavior analysis to enhance recommendation relevance, engagement, and system responsiveness in online platforms. This study employed a quantitative, applied research design with a developmental orientation. The population consisted of active users of large-scale online platforms in Tehran, selected based on continuous platform usage and availability of real-time interaction data. Behavioral data were collected through system-level logging mechanisms that captured fine-grained, time-stamped interaction signals, including click behavior, dwell time, navigation patterns, and session dynamics. These data were complemented by contextual indicators related to usage conditions. The proposed framework integrated real-time analytics, sequential behavior modeling, and adaptive learning mechanisms to generate recommendations dynamically during active sessions. Data preprocessing, feature engineering, and model training were conducted to support continuous inference under real-world latency constraints. Inferential analyses indicated that the proposed real-time intelligent framework significantly outperformed rule-based, collaborative filtering, and static machine learning models across key performance indicators. The framework achieved higher click-through rates, longer engagement durations, and greater recommendation acceptance, reflecting improved alignment with users’ immediate preferences. Real-time model adaptation following short-term behavioral changes led to statistically meaningful increases in session depth and reductions in bounce behavior. High short-term preference prediction accuracy was achieved alongside low inference latency, confirming the framework’s effectiveness and operational feasibility in dynamic online environments. The findings demonstrate that incorporating momentary user behavior analysis into an adaptive real-time recommendation framework substantially improves personalization quality and user engagement, supporting the shift toward behavior-aware intelligent recommender systems in modern online platforms.</p> https://www.bmfopen.com/index.php/bmfopen/article/download/377/283