利用AI技術偵測假新聞之實證研究

Empirical Research on Fake News Detection Using AI Technology

王維菁(Wei-Ching Wang) ； 廖執善(Chih-Shan Liao) ； 蔣旭政(Hsu-Cheng Chiang) ； 周昆璋(Kun-Chang Chou)

《中華傳播學刊》 39期 (2021/06) Pp. 43-70

https://doi.org/10.3966/172635812021060039002

摘要

數位媒體科技日新月異，為人類創造無比便利的溝通工具，卻也讓我們身處於假新聞與假訊息充斥的環境。為了因應假新聞的氾濫，本研究運用人工智慧（Artificial Intelligence, AI）與自然語言處理NLP技術，結合深度學習方法，設計出一套新聞可信度評估系統模型。本系統經測試後辨識率最高可達90.53%。然AI決策的可解釋性、人類自主性與社會過度依賴等問題皆是運用AI於假新聞上需深思者。本系統除了作為網路使用者在評估新聞是否真實可信之輔助工具，也期盼提醒民眾對於假新聞有所警戒。

關鍵字

人工智慧 ； 自然語言處理 ； 社群網路 ； 假新聞

並列摘要

Digital media technology is progressing rapidly, providing convenient communication tools for mankind. However, it is also exposing people to volumes of fake news and messages. In response to the flood of fake news, the authors of this study designed a news credibility system model that combines artificial intelligence (AI) technology, natural language processing, and deep learning to perform multidimensional and multiangle data analyses. Our system reached 90.53% in the confidence index for identifying unverified news. However, the decision interpretability of AI, human autonomy, and social overdependence are all points to consider when using AI to detect fake news. AI can be used as an effective auxiliary tool by Internet users to verify news credibility and can hopefully prompt vigilance against fake news.

並列關鍵字

artificial intelligence ； natural language processing ； social media ； fake news

參考文獻

Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146-1151 doi: 10.1126/science.app9559

何吉森（2018）。〈假新聞之監理與治理探討〉，《傳播研究與實踐》，8(2)：1-41。

Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146-1151.

Wardle, C. (2017, February 16). Fake news. It’s complicated. First Draft. Retrieved from https://firstdraftnews.com/fake-news-complicated/

何吉森（2018）。〈假新聞之監理與治理探討〉，《傳播研究與實踐》，8(2)：1-41。