期刊文献+

国内主流搜索引擎的算法审计研究 被引量:9

An Algorithmic Auditing of Mainstream Search Engines
原文传递
导出
摘要 在数字信息生态中,商业化搜索引擎和商业性信息网站分别扮演着日益重要的角色。本研究采用算法审计方法,以“新冠病毒”这一公共议题为例,对国内主流搜索引擎的检索结果进行审计,基于大数据和图计算识别二者间的商业关联,并计算上述关联关系是否导致搜索引擎信息过滤、排序的偏向或降低搜索结果的多样性。研究证实,商业关联会导致搜索引擎算法偏向,即搜索引擎首先倾向于呈现来自相同利益主体网站的信息,其次是利益相关主体网站的信息,并赋予二者更优先的排序;利益关联对内容多样性的影响因搜索引擎而异。研究同时发现,在目前的互联网信息检索服务中,商业信息网站处于“内容提供商”和“广告商”之间的模糊地带;搜索引擎和商业信息网站之间基于商业关联形成的“渠道—内容”合谋构成了对数字生态公共性的挑战。 Commercial search engines and commercial information websites are serving as increasingly important channels and content suppliers respectively in online digital environment. Using the algorithmic auditing as a method, this study explores whether the commercial connections between the two entities systematically bring about decision biases of search engine algorithms. Taking the contents about the Covid-19 as an example, the research audits the results of four popular search engines in China, including Baidu, Sogou, Haosou, Shenma, about their representation of contents from websites such as Tencent, Net Ease, Sina, Sohu, Phoenix, and Zhihu. We identify the commercial connections between the search engines and websites with big data graph analysis, and calculate whether the above-mentioned connections lead to search engine bias in terms of information filtering and sorting, and content diversity. Studies have confirmed that commercial associations can lead to a bias in search engine algorithms, that is, search engines tend to present information from the same-stakeholder websites first, followed by information from websites belonging to their commercial partners, and give them higher priority. The impact on content diversity varies by search engine. On the basis of the above findings, the research further discusses search engine’s bias driven by business motivation and its potential influence. We argued that the role of commercial websites is blurred between content provider and advertiser;the reciprocal relationship between search engines and commercial websites posts threat to the publicity of digital information ecology.
作者 师文 陈昌凤 SHI Wen;CHEN Chang-feng
出处 《新闻大学》 CSSCI 北大核心 2022年第10期84-100,122,123,共19页 Journalism Research
基金 国家社科基金重大项目“智能时代的信息价值观引领研究”(18ZDA307)。
关键词 搜索引擎 算法审计 算法伦理 search engine algorithm audit algorithm ethics
  • 相关文献

共引文献6

同被引文献124

引证文献9

二级引证文献32

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部