This paper proposed an innovative method based on the semantic differential of Kansei Engineering and the correlation analysis of machine learning in the case of small-size samples, which could achieve the evaluation of users’ subjective feelings to form factors of modeling language and color schemes in the process of product design. It used grey relational analysis and the entropy method to analyze and study the design of modeling and color of guide signs in a public space, and converted the subjective assessment process of designer and user in art design into numeric operation that quantifies and treats data. This research faced two problems: one was to analyze the form correlation between the modeling and color of designed guide signs and the architecture and environment of a public space where the signs are placed, and the other was to convert the subjective assessment of designed guide signs into objective quantified data. The paper took vertical guide boards in a guide sign system as a research object, analyzed and resolved it according to the primitive type of form factors of modeling language, and converted them into numbered sequences of the form factor type. Then, guide boards using different color schemes were utilized to conduct group tests on users; the uniformity of guide boards and architectural style was evaluated through a semantic differential method of Kansei engineering; the test data was analyzed and treated using the grey relational degree, Pearson correlation coefficient and the entropy method; the user’s qualitative assessment on the correlation degree of the modeling form and architectural style of guide boards using different color schemes was converted into quantitative numerical parameters; and finally, we obtained the assessment results of the correlation degree and significance of integrated color and modeling language form factors. Specific to the users’ assessment discrepancies on the style design of guide signs using different colors, this paper proposed a color scheme evaluation approach with uniform color space based on the grey correlation degree, used Pearson correlation coefficient to make a numerical analysis on users’ subjective assessment, presented a modified entropy method based on standardization, and made a comprehensive assessment on the uniformity of style design in different colors and architectural style. Thus, the numerical analysis results of the uniformity between the guide signs in different color schemes and modeling forms and the architectural style of public spaces were obtained. This research method can provide a new research means and numerical computation method to the study of color and form correlation in product design and style design. Meanwhile, it also provides a theoretical foundation and application method for artistic design by virtue of high-performance computing platforms.
