<p>The electrical conductivity of CaO-based slag systems significantly influences various aspects of their metallurgy. The existing optical basicity model has a limited application range and involves complex parameter fitting. Additionally, the geometric model is restricted to ternary slag systems. In response to these limitations, this paper proposes a machine learning-assisted model for refining the estimation of optical basicity in relation to the electrical conductivity of CaO-based slag systems. A comprehensive public database has been developed encompassing CaO-SiO₂-Al₂O₃, CaO-SiO₂-Al₂O₃-MgO, and CaO-SiO₂-Al₂O₃-MgO-La₂O₃ slag systems, which has been subsequently utilized for model validation. The results indicate that this approach substantially enhances both the accuracy and generalization capability of the model. Furthermore, it introduces a novel perspective on data analysis and mechanism-model coupling in exploring the physicochemical properties of slag systems.</p>