Classification of Hashtag's Popularity Growth Patterns Using Shape Dynamic Time Warping

This study presents a new method for classifying growth patterns of cultural elements, specifically focusing on hashtag popularity in social media. We use the Yule-Simon process as a baseline and apply Shape Dynamic Time Warping (ShapeDTW) to categorize hashtags into distinct growth patterns. Our approach successfully identified typical growth patterns by setting similarity thresholds. Notably, our linear and logarithmic scale comparison revealed that the latter was more effective in identifying characteristic growth patterns. This was particularly evident for time series demonstrating Preferential Attachment-like mechanisms. Our findings highlight the importance of considering both scale and underlying growth mechanisms when analyzing time series of cultural phenomena.