Abstract
This study addresses the prevalent challenge of experience-driven operational decision-making in shooter games on the Steam platform. Focusing on four representative titles—Counter-Strike 2, War Thunder, Team Fortress 2, and PUBG—we systematically collected multi-dimensional time-series data (historical reviews, concurrent player counts, user engagement metrics) and leveraged NLP techniques to precisely identify and timestamp operational events such as discounts and content updates. A composite popularity index V(t) was constructed via the entropy weight method, integrating review activity, player concurrency, sentiment-adjusted ratings, and retention indicators. A differential equa- tion model incorporating memory effects, word-of-mouth modulation, event-driven impulses, and natural decay was then established; rigorous validation (mean R² = 0.91, RMSE < 8.5%, peak coincidence rate > 85%) confirmed its high fidelity in capturing dynamic popularity evolution. Building on this foundation, an integer programming model was formulated to maximize cumulative popularity gains over a 12-month horizon by optimizing discount magnitudes (5%–20%) and content update scheduling under operational constraints. Simulations revealed the optimized strategy elevates average popularity by 18.3% across titles while mitigating user fatigue and(popularity depletion) risks. Robust- ness was further verified through Monte Carlo simulations with ±20% parameter perturbations: the strategy maintained performance stability (fluctuations < 5% at 95% confidence interval), significantly outperforming conventional approaches. The resulting transferable, data-driven decision framework not only delivers actionable scheduling insights for shooter games but also provides a scalable methodology for RPG, MOBA, and other genres, empowering the gaming industry to transition from intuition-based practices to intelligent, evidence-informed operational strategy.