Narrative Landscape: Mapping Narrative Dispositions Across LLMs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Jung, Donghoon, Choi, Jiwoo, Chae, Songeun, Jung, Seohyon
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909029560745984
author Jung, Donghoon
Choi, Jiwoo
Chae, Songeun
Jung, Seohyon
author_facet Jung, Donghoon
Choi, Jiwoo
Chae, Songeun
Jung, Seohyon
contents This study proposes a quantitative framework for profiling LLM dispositions as stable, model-specific regularities in output under repeated, controlled elicitation. Using a structured narrative constraint-selection task administered across six frontier models and three instruction types, we operationalize disposition through two dimensions: "consistency", measured as cross-replication selection overlap via Jaccard similarity, and "diversity", measured as dispersion across options via the inverse Simpson index. We further introduce Narrative Landscape, a PCA-based visualization that maps each model's selection profile into a shared space for direct comparison. Results reveal a clear rigidity-exploration spectrum across model families and show that instruction types shift the geometry of selection spaces even when scalar metrics appear similar, indicating that comparable scores can mask qualitatively distinct selection topologies.
format Preprint
id arxiv_https___arxiv_org_abs_2605_08742
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Narrative Landscape: Mapping Narrative Dispositions Across LLMs
Jung, Donghoon
Choi, Jiwoo
Chae, Songeun
Jung, Seohyon
Computation and Language
Artificial Intelligence
This study proposes a quantitative framework for profiling LLM dispositions as stable, model-specific regularities in output under repeated, controlled elicitation. Using a structured narrative constraint-selection task administered across six frontier models and three instruction types, we operationalize disposition through two dimensions: "consistency", measured as cross-replication selection overlap via Jaccard similarity, and "diversity", measured as dispersion across options via the inverse Simpson index. We further introduce Narrative Landscape, a PCA-based visualization that maps each model's selection profile into a shared space for direct comparison. Results reveal a clear rigidity-exploration spectrum across model families and show that instruction types shift the geometry of selection spaces even when scalar metrics appear similar, indicating that comparable scores can mask qualitatively distinct selection topologies.
title Narrative Landscape: Mapping Narrative Dispositions Across LLMs
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2605.08742