IMPACTeen: Intentions, Manipulation, Persuasion, Annotations, and Consequences in Teen Communication Dataset

IMPACTeen is a newly released dataset comprising 1,021 texts and 5,100 annotations focused on social influence scenarios among adolescents, annotated from five perspectives including teenagers and psychologists. It was generated using constrained LLM techniques and underwent a two-step human editing process to enhance realism. This dataset is significant for practitioners as it facilitates research in social influence detection and cross-lingual modeling, providing a resource for training and evaluating language models in youth communication contexts.