The dynamic behavior of assembled structures is strongly influenced by the dynamic properties of connections between the individual substructures, commonly referred to as joints. Accurately predicting assembly dynamics relies on identifying joint properties, which are influenced by factors such as preload, temperature, and vibration amplitude. These interactions make analytical and numerical modeling challenging, necessitating an experimental modeling approach. This study presents a joint identification framework inspired by the Lagrange multiplier frequency-based substructuring, along with four mutually independent modifications to the identification approach. These modifications comprise sparse regression, rigid-body mode constraints, excitation direction updating, and indirect parametrization. A parametrization-based approach is employed for estimating the mass, damping, and stiffness properties of the joint. The numerical and experimental results highlight the benefits of each modification and demonstrate the effectiveness of the proposed framework for robust joint property estimation.