Handling uncertainty is an important factor in decision-making. To handle uncertainty, various methodologies and approaches have been introduced using fuzzy sets and its extensions. However, these are limited to the structural space bounds in representing the information given by the experts or decision-makers. On the other hand, linear diophantine fuzzy sets (LDFSs) eradicate this limitation of the fuzzy set theories and handle uncertainty even more precisely with the inclusion of reference parameters. In this paper, a consistent multi-attribute decision-making (MADM) approach is proposed based on linear diophantine fuzzy soft set (LDFSSs) and revised max-min average composition method. Firstly, we propose LDFSS as a hybridization of LDFS and soft set, and then we use it in the revised max-min average composition method. Linear diophantine fuzzy soft matrix (LDFSM) is used to represent the evaluating values of the alternatives corresponding to the attributes. Next, the LDFSMs given by various decision-makers are aggregated using the max-min average composition method. Then the score values of the alternatives are computed from the aggregated LDFSM. An alternative with the maximum score value is selected as the better one. Finally, applicability and consistency of the proposed approach have been illustrated using practical example.