- 期刊
網民如何看待ChatGPT——基於B站視頻的彈幕文本分析
How do netizens view ChatGPT - bullet screen text analysis based on Bilibili videos
摘要
基於B站視頻海量評論數據進行文本分析，瞭解網民對ChatGPT這種新技術的情感、對新技術的期盼，集思廣益，對確立中國大陸類人工智慧的發展方向，改進類人工智慧技術具有重要意義。【方法／過程】通過python網路爬蟲、語義網路特徵關聯分析、詞雲圖、情感分析方法，對基於B站ChatGPT相關視頻彈幕文本進行了情感計算與分析。【結果／結論】各類視頻彈幕高頻詞都出現了較為明顯的分組現象，網民在這三類視頻中表現出的情感分佈相對穩定，中性情感占多沒有特別大的情感波動，主要話題集中在對人類發展方向的思考、AI人工智慧的思考等。【創新／局限】雖然數據來源於現實世界，但基於樣本量範圍、電腦演算法的情感分析並不能完全準確反映網民的傾向。
並列摘要
Based on the massive review data of Bilibili video, text analysis is conducted to understand the emotions and expectations of netizens towards ChatGPT, a new technology, and gather ideas. This is of great significance for establishing the development direction of artificial intelligence in China and improving artificial intelligence technology. [Method/Process] Through Python web crawler, semantic network feature correlation analysis, word cloud graph, and sentiment analysis methods, sentiment calculation and analysis were conducted on video barrage texts related to ChatGPT on Bilibili. [Result/Conclusion] There is a significant grouping phenomenon of high-frequency words in various types of video barrage. The emotional distribution of netizens in these three types of videos is relatively stable, with a majority of neutral emotions without significant emotional fluctuations. The main topics focus on thinking about human development direction, AI artificial intelligence, and so on. Although the data comes from the real world, sentiment analysis based on sample size range and computer algorithms cannot fully accurately reflect the tendencies of netizens.