Adaptive Text Generation with Personality Types and Continuous Emotion Intensity Dataset

This dataset is released as part of the study:Adaptive Text Generation with Personality Types and Continuous Emotion IntensityIt contains a Chinese controllable text generation dataset integrating MBTI personality types and continuous emotion intensity annotations, designed for research in affective natural language generation. Dataset OverviewTo ensure consistency and usability, the generated texts are constrained to a length of 10–50 Chinese characters.

For each MBTI personality type, 15,000 samples are generated, resulting in a balanced dataset across all 16 personality categories. Each sample is further annotated with a six-dimensional emotion intensity vector using an external emotion annotation model.formatted_style_data.csv (240000 samples) File FormatEach row contains the following fields:text,angry,fear,happy,neutral,sad,surprise,type,ME,target_text Task DefinitionThis dataset supports the task of:Controllable Text Generation conditioned on Personality Type and Continuous Emotion IntensityThe model must generate fluent and coherent text that:Aligns with the specified MBTI personality typeReflects the given multi-dimensional emotion intensity distributionPreserves semantic consistency with the original text Emotion RepresentationEmotion is represented as a continuous probability-like distribution over six basic emotions:Emotion Descriptionangry anger intensityfear fear intensityhappy happiness intensityneutral neutralitysad sadnesssurprise surprise intensityThe sum of values is not constrained to 1, reflecting multi-label emotional expression.