YouTube 趨勢模型:以文字探勘分析聚焦電競產業
No Thumbnail Available
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
隨著電競產業的快速成長與數位行銷的融合,了解YouTube網紅行銷成功的關鍵因素變得極為重要。本研究採用文字探勘技術,聚焦於電競產業,透過分析YouTube影片留言,深入解析觀眾的真實反饋。本研究蒐集了1,038部高點閱率影片,其中包含電競品牌與網紅合作之影片,目的在揭露傳統量化指標未能捕捉的情感與主題洞察。研究方法包含網路爬蟲(使用 Selenium)、文本前處理、詞頻分析、潛在狄利克雷分配(LDA)主題建模,以及透過NRC情緒詞典進行情感分析。情緒分析結果顯示,贊助影片與非贊助影片在正面與負面情緒反應上存在顯著差異。此外,LDA主題模型分析揭露觀眾特別關注產品實用性、效能規格、性價比與品牌認知等主題,提供了市場策略擬定的細緻參考。
本研究發現,儘管贊助標示對觀眾情緒影響有限,但產品真實性、評測者可信度及實際使用情境等因素顯著影響觀眾情緒與認知。因此,本研究建議電競品牌在網紅合作影片中,應強調真實的使用體驗與詳細的產品效能展示,以提升消費者信任與情感連結。本研究結合文本分析與行銷實務策略,期望提供電競行銷人員更深入的洞察,以精進網紅合作策略、優化內容創作,並有效配置行銷資源。
With the rapid growth of the e-sports industry and its integration with digital marketing, understanding the drivers behind successful YouTube influencers campaigns has become critically important. This study employs text mining techniques, focusing specifically on the esports sector, to analyze and interpret audience engagement through YouTube video comments. Using a dataset comprising 1,038 highly-viewed videos, including collaborations between esports brands and influencers, this research aims to uncover the hidden emotional and thematic insights that traditional quantitative metrics fail to capture.A comprehensive methodological framework was developed involving web scraping (Python Selenium), text preprocessing, frequency analysis, topic modeling (Latent Dirichlet Allocation, LDA), and sentiment analysis using the NRC Emotion Lexicon. The sentiment analysis reveals significant emotional differences between sponsored and non-sponsored videos, particularly highlighting variations in positive and negative audience reactions. Furthermore, the LDA model identifies distinct thematic concernsamong audiences, such as product usability, performance specifications, price-value considerations, and brand perception, offering nuanced insights for strategic marketing decisions.The findings suggest that while sponsorship disclosure does impact audience sentiment slightly, broader factors such as product authenticity, reviewer credibility, and real-life application significantly influence audience perceptions and emotional engagement. Consequently, this study recommends esports brands emphasize authentic experiences and detailed performance demonstrations within influencer campaigns to enhance consumer trust and emotional resonance. By bridging textual analysis with practical marketing strategies, this research contributes valuable insights for esports marketers aiming to refine their influencer collaboration approaches, optimize content creation, and strategically allocate marketing resources.
With the rapid growth of the e-sports industry and its integration with digital marketing, understanding the drivers behind successful YouTube influencers campaigns has become critically important. This study employs text mining techniques, focusing specifically on the esports sector, to analyze and interpret audience engagement through YouTube video comments. Using a dataset comprising 1,038 highly-viewed videos, including collaborations between esports brands and influencers, this research aims to uncover the hidden emotional and thematic insights that traditional quantitative metrics fail to capture.A comprehensive methodological framework was developed involving web scraping (Python Selenium), text preprocessing, frequency analysis, topic modeling (Latent Dirichlet Allocation, LDA), and sentiment analysis using the NRC Emotion Lexicon. The sentiment analysis reveals significant emotional differences between sponsored and non-sponsored videos, particularly highlighting variations in positive and negative audience reactions. Furthermore, the LDA model identifies distinct thematic concernsamong audiences, such as product usability, performance specifications, price-value considerations, and brand perception, offering nuanced insights for strategic marketing decisions.The findings suggest that while sponsorship disclosure does impact audience sentiment slightly, broader factors such as product authenticity, reviewer credibility, and real-life application significantly influence audience perceptions and emotional engagement. Consequently, this study recommends esports brands emphasize authentic experiences and detailed performance demonstrations within influencer campaigns to enhance consumer trust and emotional resonance. By bridging textual analysis with practical marketing strategies, this research contributes valuable insights for esports marketers aiming to refine their influencer collaboration approaches, optimize content creation, and strategically allocate marketing resources.
Description
Keywords
YouTube, 文字探勘, 情緒分析, 主題模型, 機器學習, YouTube, Text Mining, Sentiment Analysis, Topic Modeling, Machine learning