Relational Semantic Structure: A Framework for Meaning Beyond Vector Similarity

Read full text Summarise with VIRAT
Authors: Jung Jae Hoon
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Current AI systems treat meaning as a fixed representation encoded in vector space and retrieved through similarity-based mechanisms. This approach captures surface-level patterns, but structurally excludes the relational, contextual, and temporal dimensions through which meaning is actually formed. This paper proposes Relational Semantic Structure (RSS) as a framework for understanding how meaning is organized beyond static representation. We argue that meaning does not reside in nodes, but emerges from relational configurations — structured through selection, constrained by context, and reinforced through collective interaction over time. We define nodes not as fixed units of meaning, but as points of relational divergence. Relations are understood as selection mechanisms that possess direction, strength, and emergent patterns. Within this framework, meaning exists not as a discrete entity but as an uneven distribution of relational possibilities, continuously reconfigured by contextual variables and observable through temporal change. We further demonstrate that the limitations of current AI systems arise not from insufficient scale or performance, but from the structural absence of relational tracking, reinforcement mechanisms, and contextual weighting. Similarity-based retrieval cannot substitute for relational selection, and therefore cannot sustain the structural continuity through which meaning persists. This paper extends the Fluid Relational Semiotic (FRS) framework introduced in previous work, shifting from the declaration that meaning is a process to a structural account of what that process produces.

Fields of Study & Subjects

Topic Modeling Cognitive Computing and Networks Syntax Semantics Linguistic Variation

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN
Document type
Preprint