Synthetic Resonant Amplification Framework Dataset (SRAF-200) and Seed Corpus (N=30)

Synthetic Resonant Amplification Framework Dataset (SRAF-200) and Seed Corpus (N=30) This repository contains the corpora accompanying the preprint Interrupting Resonant Amplification: A Mechanistic and Design Framework for Human–AI Interaction (Kim, 2025, DOI: 10.5281/zenodo.17019212). It provides both a small naturalistic seed sample and a synthetic dataset for measurement development, annotation, and reproducibility. Contents RAF_MiniDataset_30entries_CleanEnglish.xlsx – A 30-entry seed set of public, de-identified, paraphrased excerpts of user–AI interactions. – Fields: User_Prompt, AI_Response, RAF_Phase, Distortion, Mechanism, Notes. SRAF_200_with_RAF_Phase_filled.csv – A fully synthetic corpus of 200 entries generated via factorial design. – Factors: RAF Phase (Attachment, Co-Creation, Linguistic Reinforcement, Internalization), Domain, Persona, Prompt Polarity, Turn Length. – Includes edge/boundary cases (+40) in addition to balanced factorial cells (160). SRAF_200_FINAL_metadata.csv – Extended metadata version of the synthetic corpus, with derived features such as inclusive-pronoun counts, lexical accommodation, and reciprocity markers (PCCI components). Purpose Grounding theoretical constructs of the Resonant Amplification Framework (RAF) in linguistic material. Developing and validating measurement indices such as the Parasocial Co-Creation Index (PCCI). Providing reproducible resources for annotation reliability and feature engineering. Ethical Note Seed set (N=30): Public, anonymized, and paraphrased excerpts; no identifiable private content. SRAF-200 (N=200): Fully synthetic; no human subjects involved. As such, no IRB approval was required. Citation: Kim, R. S. (2025). Synthetic Resonant Amplification Framework Dataset (SRAF-200) and Seed Corpus (N=30). Zenodo. 10.5281/zenodo.17019247