基於社交媒體資料的運動員熱度波動分析——以全紅嬋2024年微博資料為例
研 究 生：李 琳
指導教授：劉先翔
摘要
目的：本研究旨在探討中國跳水運動員全紅嬋在2024年整年度微博上的社交媒體熱度變化，並分析這些變化與微博內容情感之間的關係。透過對比賽前後的數據分析，研究試圖揭示社交媒體熱度的波動規律，以及不同情感內容對粉絲互動的影響，以提供運動員社交媒體熱度波動的深層次解釋，並理解其背後的機制。方法：本研究運用時間序列分析（ARIMA模型）與文本情感分析（LSTM模型），對全紅嬋的微博發文數據、粉絲互動量及文本內容進行量化分析。透過數據整理與描述性統計，識別微博互動趨勢，並利用情感分析技術分類微博內容，進一步探討不同情感類型對互動熱度的影響。結果:研究結果顯示，全紅嬋的微博互動量在比賽前後呈現明顯變動，比賽當日達到最高峰，並於賽後數日內逐漸回落。此外，情感分析結果表明，正向情感內容（如比賽勝利、粉絲回應）能顯著提升互動量，且影響時間較長，而負向情感內容（如爭議、批評）雖能短期內引發關注，但熱度下降速度較快。結論:本研究證實了比賽前後的社交媒體熱度變化具有可預測性，且微博內容的情感傾向對互動量有顯著影響。研究結果可為運動員及其團隊在社交媒體運營方面提供指導，幫助制定更有效的內容策略。此外，未來研究可進一步結合影像與影片數據，探索多模態信息在社交媒體影響力中的作用。

Analysis of Athlete Popularity Fluctuations Based on Social Media Data: A Case Study of Quan Hongchan's Weibo Data in 2024
Student：Li Lin
Advisor：Liu, Hsien-Hsiang
Chang, Chun-Yi
Abstract
Objective: This study investigates the fluctuations in social media heat surrounding Chinese diver Quan Hongchan on Weibo throughout the year 2024. It aims to examine how changes in online popularity relate to the emotional tone of her microblog content. By analyzing user engagement before and after competitions, the research seeks to uncover patterns in attention dynamics and explore how different emotional expressions influence fan interaction.
Methods: The study adopts time series analysis using the ARIMA model and sentiment analysis based on an LSTM model to quantitatively assess Weibo post volume, interaction metrics, and textual emotions. Descriptive statistics are applied to identify trends in engagement heat, while sentiment classification is used to explore how various emotional tones affect the intensity and duration of audience attention.
Results: The findings reveal that Quan Hongchan’ s social media heat levels showed significant fluctuations around competition periods, peaking on the day of the event and gradually declining in the days that followed. Positive emotional content such as competition victories and supportive fan responses-significantly increased interaction volume and sustained heat for a longer period. In contrast, negative content-such as controversy or criticism-drew short-term attention but experienced a rapid drop in popularity.
Conclusion: The study confirms that the fluctuations in an athlete’ s social media heat are both emotion-driven and time-sensitive. The emotional tone of microblog content plays a crucial role in determining engagement intensity and retention. These insights offer practical value for athletes and their media teams in crafting emotionally strategic content to maintain and boost online popularity. Future research could integrate multimodal data such as images and videos to explore how visual elements further affect social media heat.