Resource Allocation Problems (RAPs) arise across domains such as healthcare, logistics, education, energy, sustainability etc.. and disciplines such as computer science, operations research, biology, and physics. Traditionally formulated as optimization or search problems, they often overlook how feasible solutions are constructed under constraints, uncertainty, and limited resources.

This thesis introduces the Unified RAP Framework, which reconceptualizes resource allocation as a constructive process within hierarchical complex systems. It models solutions as structured transformations governed by constraints and resource availability, and introduces two key concepts: "Constructibility", capturing the effort of building solutions, and Solution Attainability Probability, measuring the likelihood of reaching a solution under bounded rationality.

The framework enables comparison of problem complexity across domains, providing a novel basis for decision support and governance in complex, multi-stakeholder environments. It also incorporates human-in-the-loop mechanisms, where diagnostic states transform infeasibility into actionable insights, supporting adaptive refinement of goals and constraints.

By connecting to theories from physics, biology, and machine learning, the thesis shows that diverse systems share common principles based on the interplay between cost and reachability. Implementations demonstrate practical applications in constraint programming and adaptive scheduling.

Overall, this work establishes a new foundation for analyzing, comparing, and governing complex resource allocation problems across domains.