Intelligent classification and dynamic evolution simulation study on air conditioner product demand characteristics

Extracting actionable user insights from online reviews is often compromised by high data noise and static analysis limitations. To address this, this study proposes an integrated framework coupling adaptive noise filtration with multi-dimensional dynamic simulation. Specifically, an Improved Whale Optimization Algorithm (IWOA) tunes Support Vector Machine (SVM) hyperparameters to ensure high-precision review classification. Subsequently, BERTopic extracts demand topics, and temporal cosine similarity tracks their evolution within a dynamic Importance–Performance Analysis (IPA) framework. Simulation reveals that air-conditioner user demands exhibit distinct seasonal nonlinear fluctuations. This quantifiable tool supports demand forecasting and strategic product iteration in the home-appliance industry.