信息诊断系统设计思路:人工核查、广众参与和人工智能的三合一运用  

Theorizing an Information Diagnosis System:The Hybrid Uses of Manual Fact-checking,Crowdsourcing and Artificial Intelligence

作　　者：曾姿颖 黄煜[2] 张引 宋韵雅[3] 周琳 Stephanie Jean Tsang;Yu Huang;Yin Nick Zhang;Yunya Céline Song;Lin Zhou(School of communication,Hong Kong Baptist University)

机构地区：[1]香港浸会大学传播系 [2]香港浸会大学传理学院 [3]香港浸会大学新闻系

出　　处：《全球传媒学刊》2021年第1期35-62,共28页Global Journal of Media Studies

摘　　要：当前对虚假新闻治理的研究较少出于技术逻辑的思路。真假新闻的界限在实践中通常微妙以及难以辨别,加上人们对真假新闻往往有着不同的见解和解读。本文基于来自香港的实证数据分析以及本地虚假新闻核查平台的初步实践经验,提出“人工核查—广众参与—人工智能”的协同核查模式,以最大限度优化社交平台虚假新闻的治理效果。The emergence of social media has bred the dissemination of misinformation, which has attracted much scholarly attention. While the literature has mainly focused on the production, dissemination and reception of(mis)information, there has been a lack of a comprehensive theoretical framework that sheds lights on how readers process misinformation as well as the absence of a practical framework to diagnose misinformation. This article reviews the current literature on misinformation and proposes an information diagnosis system by answering the following three research questions.(1) What causes a person to question the authenticity of information and the author’s intentions?(2) How would a person verify a piece of suspicious information?(3) How should a fact-checking agency conduct verification to meet the challenges posed by the above two research questions? Two rounds of survey data collected in Hong Kong in 2019(N=1211) and 2020(N=1223) revealed that audiences held a diversity of definitions when it comes to misinformation and that they would only verify information perceived as suspicious. In addition, using HKBU FactCheck Service, a local fact-checking center, as a case study, we propose a hybrid information diagnosis system to verify misinformation to overcome the limitations of traditional verification methods. Given the rapid development of technologies, such as data mining, machine learning, and deep learning, it is necessary to take advantage of automation to complement the work of fact-checkers in misinformation detection. Hybrid applications of manual fact-checking, crowdsourcing, and artificial intelligence suggest the need to pursue fact-checking in a collaborative, interdisciplinary manner.

关 键 词：虚假新闻 信息诊断 人工智能 事实核查 

分 类 号：G219.2[文化科学—新闻学]