[总结翻译] Using eye movement to improve language processing neural networks 用眼部活动来提高语言处理模型的功能

new source 新闻来源：https://wiredgorilla.com/tracking-readers-eye-movements-can-help-computers-learn/

ETH Zurich graduate student Nora Hollenstein is applying readers' eye movements to language processing neural networks. The idea is that the eye movements while reading reveal what happens in our brains. For instance, by using the information on how long the eyes stay on each word, Hollenstein and her collaborators help neural networks focus on the most important parts of a sentence. The team has already seen some results in studying sentiment and finding hate speech, grammatical errors and special names. However, it's hard to gather useful gaze information at the moment.

ETH Zurich 研究生 Nora Hollenstein 把读者的眼部活动数据用到语言处理的神经网络里。出发点是我们阅读时的眼部活动可以反应出我们的大脑活动。Hollenstein和她的合作者的做法之一，是利用人眼在每个词上停留的时间来让神经网络专注于句子中重要的部分。她们的团队已经在分析情绪，以及识别仇恨言辞，语法错误和专属名词看到了一些成果。不过目前有用的眼光信息是不容易被收集到的。

图片发自App