The existing liquefaction evaluation model inevitably contains uncertainty in the establishment process, which will affect the evaluation of the effect of seismic liquefaction. The purpose of this study is to combine cone penetration test and shear wave velocity test (CPT-Vs) to develop a simplified probabilistic framework for liquefaction potential assessment, and to couple the first-order reliability method (FORM) to reduce model and parameter uncertainties. Based on the case histories of CPT-Vs, the framework uses the Bayesian-optimized CatBoost algorithm to develop a limit state function for liquefaction trigger analysis. Within the framework of FORM, this limit state function parameter and model uncertainties are characterized by the mean (μc) and coefficient of variation (COV) of the soil parameters, which are obtained and calibrated by combining site-specific new information with prior knowledge. Moreover, in the absence of parameter uncertainty, a deterministic model is developed which establishes a mapping function between probability of liquefaction to factor of safety. The study results show that the uncertainty of the limit state function model can be characterized by using μc=1.14 and COV=0.30, and the simplified liquefaction probabilistic framework can provide reasonable liquefaction evaluation results.