Who did what to whom? Syntactic parsing in man and machine

谁对谁做了什么？

• 批准号: 1898610
• 金额: --
• 依托单位: University of Birmingham
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2017
• 资助国家: 英国
• 起止时间: 2017 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-1898610
• 关键词: Who; did; what; whom; Syntactic

This industrial collaboration project with Google Research London bridges between artificial intelligence, computational linguistics, computational and cognitive neuroscience. Language processing is one of the most complex tasks facing the human brain in everyday life. In order to understand 'who does what to whom', a biological or artificial agent needs to assign a syntactic structure to each sentence - a process coined 'syntactic parsing'. Artificial intelligence and computational linguistics has developed a range of machine learning algorithms (e.g. dependency parsing) that are trained on large annotated linguistic corpora to assign grammatical classes and syntactic structures automatically to novel natural sentences. Yet, while these algorithms are relatively efficient, they do not obtain the accuracy of human readers. This interdisciplinary project will combine expertise from human neuroscience (University of Birmingham) and computational linguistics (Google Research London) to determine the neural mechanisms underlying sentence comprehension in the human brain and advance parsing algorithms in machines. To study natural language processing and the underlying neural mechanisms in humans, we will measure eye movements, behavioural (psychophysics) and electrophysiological responses (EEG/fMRI) in participants reading natural sentences from syntactically annotated corpora. We will employ advanced machine learning algorithms to characterize the computational operations and neural mechanisms underlying syntactic processing in the human brain. Conversely, the insights obtained from human neuroimaging (EEG/fMRI) and eye tracking will provide critical constraints on the parameters and algorithms used in machine.

这个与谷歌研究伦敦的工业合作项目在人工智能，计算语言学，计算和认知神经科学之间架起了桥梁。语言处理是人类大脑在日常生活中面临的最复杂的任务之一。为了理解“谁对谁做了什么”，生物或人工智能体需要为每个句子分配一个句法结构——这个过程被称为“句法解析”。人工智能和计算语言学已经开发了一系列机器学习算法（例如依赖解析），这些算法在大型注释语言语料库上进行训练，自动为新的自然句子分配语法类和句法结构。然而，虽然这些算法相对有效，但它们无法获得人类读者的准确性。这个跨学科项目将结合人类神经科学（伯明翰大学）和计算语言学（谷歌研究伦敦）的专业知识，以确定人脑中句子理解的神经机制，并在机器中推进解析算法。为了研究人类的自然语言处理和潜在的神经机制，我们将测量参与者从语法注释的语料库中阅读自然句子时的眼动、行为（心理物理学）和电生理反应（EEG/fMRI）。我们将采用先进的机器学习算法来描述人脑中语法处理的计算操作和神经机制。相反，从人类神经成像（EEG/fMRI）和眼动追踪中获得的见解将为机器中使用的参数和算法提供关键约束。