An interpretable model to measure fakeness and emotion in news
An interpretable model to measure fakeness and emotion in news
复制标题
衡量新闻中的虚假性和情感的可解释模型
DOI：
10.1016/j.procs.2020.08.009
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Paul Guélorget
中科院分区：
文献类型：
--
作者：
G. Gadek;Paul Guélorget
假新闻和后真相到处都是。大量的在线新闻媒体和内容创建的频率强调了对自动信息评估工具的需求。以前的工作通常集中在自动事实检查或虚假识别上：前者试图将一段内容与可信的信息相匹配，足以证实或削弱索赔。后者收集线索，以帮助读者对内容进行评估。在这个领域，没有什么银弹：读者需要可验证的信息，因此假新闻检测器应该是可解释的。在这篇文章中，我们提出了TC-CNN：一个可解释的文本分类器。我们将其用于两个任务：假新闻检测和情感分类。第二个贡献依赖于这两个分类器，以及第三方仇恨检测器，对今年的真实的和假新闻媒体文章进行案例研究，比较主流媒体和另类右翼媒体。
Fake news and post-truth are everywhere. The huge number of online news outlets and the frequency of content creation underlines the demand for automatic information evaluation tools. Previous work usually focuses either on automatic fact-checking, or on fake-looking identification: the former tries to match a piece of content with trustable information, enough to confirm or infirm the claims. The latter gathers clues to help the reader’s assessment of the piece of content. In this domain, there is no silver bullet: the reader desires verifiable information, thus thefake news detectorshould be interpretable or explainable.In this article, we propose TC-CNN: an interpretable text classifier. We use it on two tasks: fake news detection and emotion classification. A second contribution relies on these two classifiers, and on a third-party hate detector, to perform a case study on this year real and fake-news press articles, in a comparison between mainstream and alt-right media.
影响因子：
8.5
作者：
Bulatov, Konstantin B.;Ingacheva, Anastasia S.;Gilmanov, Marat I.;Chukalina, Marina V.;Nikolaev, Dmitry P.;Arlazarov, Vladimir V.
通讯作者：
Arlazarov, Vladimir V.