HyperPersona Framework Uses Hypergraphs for Text-Based Personality Prediction

Researchers have proposed HyperPersona, a multi-level hypergraph framework for text-based automatic personality prediction (APP). The framework explicitly models the hierarchical organization of text at document, sentence, and word levels using hypergraph structures. It addresses the limitation of existing approaches that rely on shallow, sequential, or single-level representations, ignoring the multi-level structure of written language. The work is detailed in arXiv preprint 2605.17355, released in 2025. The framework treats language as a repository of socially and psychologically significant traits, aiming to infer personality from linguistic behavior as a scalable alternative to traditional psychometric assessments.