This study proposes a graph-based framework to identify prestigious members on social media platforms by analyzing their prestige evolution over time. The framework establishes behavior networks from user interactions over 10 time periods on Flickr and analyzes degree distribution, in/out degree correlation, and mixing patterns to identify currently active members. It then predicts future activity and prestige using four indicators: homophily, triadic interaction rules, continuous interests, and recency effects. Prestige is calculated based on favor volume, coverage, and timeliness. The framework extends to predicting future prestige evolution.