CAREER: Flexible Learning for Natural Language Processing

统计学习现在是自然语言处理(NLP)的核心。弥合学习和语言表达之间的差距需要超越学习参数。这个CARER项目解决了三个具有挑战性的悬而未决的问题：1.鉴于最近在学习语言模型参数和近似推理方面的进展，特征设计的过程如何能够自动化？2.考虑到NLP任务通常是在没有求助于实际应用的情况下定义的，并且特定的注释数据集不太可能满足多个NLP项目的需要，是否可以扩展学习框架来执行自动任务细化，简化语言学分析任务，以获得更一致、更精确或更快的性能？3.语言的计算模型能否考虑到我们的语言数据嵌入的非文本上下文？在最近社会文本分析和文本驱动预测的成功基础上，CAREER项目寻求利用上下文来完善语言结构的模型，同时使这一应用领域的进步成为可能。这一基础研究支持广泛的语言工程应用和离散数据分析的进步。除了核心研究进展，这个职业项目还贡献了一个新的公开可用的解析器，它对句法结构中最容易学习的元素进行建模。教育活动包括在PI的本科生自然语言处理课程中的一个基于文本驱动的预测的新项目，以及一个新的机器学习本科课程。它通过在非CS课堂上展示他的研究的各个方面，支持PI参与面向高中生和CMU更广泛的学生的外联活动。

Statistical learning is now central to natural language processing(NLP). Bridging the gap between learning and linguisticrepresentation requires going beyond learning parameters. This CAREERproject addresses three challenging, unresolved questions:1. Given recent advances in learning the parameters of linguisticmodels and in approximate inference, how can the process of featuredesign be automated?2. Given that NLP tasks are often defined without recourse to realapplications and that a specific annotated dataset is unlikely tofulfill the needs of multiple NLP projects, can learning frameworks beextended to perform automatic task refinement, simplifying alinguistic analysis task to obtain more consistent, more precise, orfaster performance?3. Can computational models of language take into account the non-textcontext in which our linguistic data are embedded? Building on recentsuccess in social text analysis and text-driven forecasting, thisCAREER project seeks to exploit context to refine models of linguisticstructure while enabling advances in this application area.This basic research supports advances in a wide range of languageengineering applications and discrete data analysis. In addition tocore research advances, this CAREER project contributes a newpublicly-available parser that models the most consistently learnableelements of syntactic struture. Educational activities include a newproject-based on text-driven forecasting within the PI's undergraduateNLP course and a new undergraduate course in machine learning. Itsupports involvement by the PI in outreach activities to high schoolstudents and to a wider range of students at CMU by exposing aspectsof his research in non-CS classrooms.