Approximate Ratio Statements in Hierarchical Weighting Models

摘要

Motivated by the difficulties of eliciting a complete preference specification,the thesis studies the representation and analysis of approximate preference statements in hierarchical weighting methods. Specifically, the thesis focuses on interval-valued ratio judgments in the modeling of partial information and develops linear programming techniques for the efficient propagation of such judgments in value tree analysis and the analytic hierarchy process. The techniques of the thesis provide intermediate dominance results after each new preference statement and thus involve the decision maker in an interactive decision support process where the set of nondominated alternatives gradually shrinks. The proposed techniques offer promising prospects for applied work and invite research into how they impact the solution of real-life problems.