Addressing the lack of theoretical guidance in screening effective alloying elements for composite interfacial alloying design，a predictive model for interfacial alloying tendencies was developed based on first-principles calculations and machine learning methods.By identifying critical characteristic parameters of alloying elements that influence heterogeneous interfacial properties，this approach accelerates the alloying design of composite materials.In this study，TiB₂/Al composites are taken as a case example，with TiB₂（0001）/Al（111） coherent and TiB₂（0001）/Al（001） semi-coherent interface models being constructed.After adding a series of alloying atoms，first-principles calculations revealed that the variation trends in the interfacial energy for the two types of interfaces were nearly opposite.Specifically，after adding Mg，Ca，Sc，Y，Zr，Ce，and Hf atoms，the interfacial energy of the coherent interface decreased significantly，whereas that of the semi-coherent interface increased substantially.Furthermore，machine learning analysis demonstrated that the variation in the interfacial energy for the coherent interface was primarily governed by the size effects of the alloy atoms （i.e.，shear modulus，Voronoi volume，and atomic radius）.Conversely，for the semi-coherent interface， the variation was dominated by the electronic effects of the alloy atoms （i.e.，work function， electronegativity，and atomic charge）.Therefore，it is revealed that the influence of alloying elements on interfacial energy primarily depends on interfacial structure and atomic properties. The impact degree of alloying elements on the performance of heterogeneous interfaces can be rapidly predicted using the Voronoi volume，shear modulus，and electronegativity of the alloy atoms.