When analyzing perceptual evaluations in industrial design, issues like limited data scope, high costs, inaccuracy, and underuse of unstructured data are common. This study introduces a novel sentiment analysis approach. Focusing on ceramic tea sets, it aims to build a model predicting their shape’s perceptual imagery. The process involves converting user reviews into quantifiable scores using sentiment analysis and identifying key design elements like handles, spouts, mouths, bellies, and lids. Quantification Theory I then maps these scores to design elements. Validated with ceramic teapot design optimizations, this method accurately captures user feelings and predicts their perceptual evaluations, aiding in design improvements and practical applications.