A Survey and Computational Atlas of Personality Models

Personality psychology has spent a century characterizing individual differences, yet computational approaches to personality are dominated by a single model: the Big Five (OCEAN). Numerous other validated frameworks exist, spanning clinical diagnosis, narcissism, motivation, cognition, conflict resolution, and applied assessment, but without a shared computational format they remain fragmented across disciplinary silos. This paper surveys 44 personality models from seven research disciplines and encodes every construct as a factor chain , which we define as a structured descriptor set that disambiguates each factor through its defining adjectives, synonyms, verbs and nouns. The result is a uniformly encoded computational atlas of factor chains paired with neural embeddings and trained classifiers, searchable across all surveyed models simultaneously. Three layers of evaluation confirm the encoding: classifiers tested against human-authored items from published psychometric instruments, an independent LLM judge panel, and replication across multiple generators, achieving 86.8% accuracy on human-authored items. Embedding-space analysis reveals that the seven disciplinary categories do not form coherent clusters in semantic space; data-driven clustering instead uncovers 15 natural groupings that cut across the predefined categories, approximate established diagnostic boundaries, and offer large-scale, lexical-level evidence consistent with personality science having studied overlapping constructs under different names for decades, though shared vocabulary alone is not sufficient evidence for construct equivalence. We position the atlas as an exploratory resource for cross-model comparison and selection rather than a replacement for validated psychometric measurement. The full atlas, trained classifiers, and reproducibility notebooks are openly released.