Abstract Chemical exfoliation of graphite is an effective method to produce graphene of relative high quality, where the choice of solvents plays an important role in the product yield and quality. Here, we performed screening of different solvents and their mixtures for the liquid-phase exfoliation of graphite under ultrasonication. A synergistic effect among aromatic, amine, and halogen groups was identified. The synergy was more effectively exploited when these functional groups were combined through solvent mixtures compared to when they were introduced in the molecular structure of single solvents. The screening results were utilized for a novel machine learning technique based on the Dempster–Shafer theory of evidence to systematically investigate synergistic effects and recommend new solvent combinations. The proposed combination of the experiment and the data-driven approach was demonstrated to be powerful for exploring synergistic solvent combinations.