Research on the construction and application of product evaluation system based on LDA and optimized FAHP model

This study addresses the limitations of traditional, expert-driven product evaluation systems by developing a scientific and objective evaluation framework that combines Latent Dirichlet Allocation (LDA) and an optimized Fuzzy Analytic Hierarchy Process (FAHP). Using Python, product attributes were extracted from e-commerce platform reviews through LDA modeling, forming the basis for constructing key evaluation indicators. By integrating FAHP with Particle Swarm Optimization (PSO), an innovative model was created to objectively determine indicator weights. The system was validated using elderly mobility aids as a case study, demonstrating a positive correlation between evaluation scores and market performance on B2B platforms. These findings confirm the framework's feasibility and scientific rigor, offering valuable insights for aligning product design with market demands and enhancing enterprise competitiveness.