This article explores the methodological significance of the synthesis of graph theory and frame semantics in modeling metaphorical meanings within cognitive linguistics. The central idea of the study is based on the description of metaphor as a multilayer conceptual network (graph), abandoning traditional lexicographic classifications. The article reinterprets the concept of metaphorical equivalence through the models of "node neighborhood" and "spreading activation". On the example of English and Uzbek, the asymmetry between the source and target domains, the functional functions of core and bridge nodes, and the semantic weight (w) of edges are analyzed. Based on the FrameNet architecture, metaphor is justified as a structural projection of one frame onto another. The results of the study show that the graph model is an effective tool for identifying cognitive disparities between agglutinative Uzbek and analytic English, creating new analytical opportunities for the fields of translation theory and natural language processing (NLP).