Frames of Feeling: The Emotional Impact of Short- vs. Long-Form Video Across Time on Social Media

This dataset comprises all Korean-language comments posted in 2023 on short- and long-form YouTube videos (each with ≥50 comments) from the top 100 Korean channels across 15 content categories. Each row includes video metadata (URL, category, format, view/like counts, season and holiday flags), the comment text with its KST timestamp, and detailed sentiment annotations (five emotion probabilities plus composite positive and negative scores). By standardizing all timestamps to KST and filtering for high-engagement videos, the data support a precise analysis of how time of day and format jointly influence consumer emotions.