首页|论新闻生产算法影响评估机制的构建

论新闻生产算法影响评估机制的构建

扫码查看
智媒时代算法已广泛嵌入新闻生产全流程,与之伴随的各类风险值得关注。当前国家层面的算法风险治理侧重具有舆论属性或社会动员能力的推荐算法和深度合成算法,并未将新闻生产相关的其他算法纳入其中,形成算法风险治理的盲区。新闻生产与公共利益密切相关,构建专门的面向新闻生产的算法影响评估机制势在必行。算法影响评估具有治理对象细分、治理节点前移、评估要素全面等优势。对新闻生产而言,算法影响评估可以提升新闻生产算法风险治理的针对性、公平分配风险责任、有效保护新闻用户权益、提高新闻从业者算法素养。新闻生产算法影响评估应根据算法风险影响对象识别风险类别、综合多重因素确定风险等级、基于风险等级确定责任义务来构建具体的评估机制。
Construction of Algorithm Impact Assessment Mechanism in News Production
The algorithm in the era of intelligent media has been widely embedded in the whole process of news production,and the various risks brought by it deserve attention.At the current national level,algorithmic risk governance focuses on recommendation algorithms and deep synthesis algorithms with public opinion attributes or social mobilization capabilities,but does not include other al-gorithms related to news production,forming a blind area of algorithmic risk governance.News production is closely related to public in-terests,so it is imperative to build a special algorithmic impact assessment mechanism for news production.The algorithm impact as-sessment has the advantages of subdivision of governance object,advance of governance node and dynamic assessment method.For news production,algorithm impact assessment can improve the pertinence of news production algorithm risk governance,fairly allocate risk responsibilities,effectively protect users'rights and interests,and improve the algorithmic literacy of news practitioners.The im-pact assessment of news production algorithm should identify the risk category according to the risk impact object of the algorithm,de-termine the risk level by combining multiple factors,and confirm the responsibility and obligation based on the risk level to build a spe-cific evaluation mechanism.

algorithm impact assessmentalgorithmic riskrisk-based approachalgorithmic literacy

张超、陈莎

展开 >

山东大学文化传播学院(山东威海 264209)

山东大学文化传播学院新闻传播研究所(山东威海 264209)

算法影响评估 算法风险 风险为本 算法素养

国家社会科学基金项目

19BXW020

2024

中州学刊
河南省社会科学院

中州学刊

CSTPCDCSSCICHSSCD北大核心
影响因子:0.854
ISSN:1003-0751
年,卷(期):2024.(2)
  • 32