an organizational constraint-aware recommender system with choice architecture-based user utility estimation and adaptive organizational strategies considering fuzzy inference

Recommender systems (rss) have emerged as indispensable mechanisms for alleviating information overload and facilitating decision-making processes. however, despite significant advances, traditional rss often overlook the cognitive processes underlying human choice. additionally, they frequently fail to consider the strategic interests of multiple stakeholders. this study proposes an adaptive approach that seamlessly integrates human decision behaviors, cognitive biases, and organizational strategies to optimize recommendation delivery. by leveraging insights from choice architecture and behavioral science, the proposed rs dynamically adjusts its recommendations to serve the overall interests of multiple stakeholders. the proposed model also simultaneously exploits the results of other rss, expert knowledge, and contextual information. the recommendation strategies, as intended by organizational experts, are modeled using a fuzzy inference system (fis). this facilitates the recommender system's ability to align the interests of both stakeholders—the organization and the customer—by integrating essential contextual data and domain-specific expertise to optimize efficiency. this approach enables rss to effectively balance individual user needs with organizational priorities. this paper evaluates the approach through a simulation-based study using kpi-driven comparisons. results show substantial improvements in organizational capacity utilization, with over-utilization decreasing from approximately 85% under the traditional rs to 19% under the proposed rs. the simulation-based validation process suggests that the proposed model can enhance organizational performance while accounting for both short- and long-term customer benefits. the results indicate that this adaptive model significantly improves overall benefits.