CI-ADDO-EN: Flexible Machine Learning for Natural Language in the MALLET Toolkit

CI-ADDO-EN：MALLET 工具包中自然语言的灵活机器学习

• 批准号: 0958392
• 负责人: Andrew McCallum
• 金额: $ 65万
• 依托单位: University of Massachusetts Amherst
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-06-01 至 2016-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0958392&HistoricalAwards=false
• 关键词: CI; ADDO; EN; Flexible; Machine

Natural language processing, information extraction, informationintegration and other text processing solutions are central componentsof computer science, and key tools for addressing the ever-increasingproblems in information overload. Issues of information overload arenot only personal problems, but critical for business productivity,national defense, and increasingly government decision-making andtransparency.State-of-the-art natural language processing is increasingly based onmachine learning. However, the methodologies can be complex, andsoftware infrastructure necessary for such systems is generallydifficult to develop from scratch. To address this need we havecreated MALLET (MAchine Learning for LanguagE) and FACTORIE (Factorgraphs, Imperative, Extensible), open-source software toolkit that runin the Java virtual machine. They provide many modernstate-of-the-art machine learning methods, specially tuned to bescalable for the idiosyncrasies of natural language data, while alsoapplying well to many other discrete non- language tasks.The project will fill three critical gaps: (1) broadening thesetoolkits' applicability to new data and tasks (with better end-userinterfaces for labeling, training and diagnostics), (2) greatlyenhancing their research-support capabilities (with infrastructure forflexibly specifying model structures), and (3) improving theirunderstandability and support (with new documentation, examples,online community support).The project will have a direct positive impact on NLP and othermachine learning research, on teaching, and on collaborative researchactivities. Well-designed toolkits not only help researchers avoidduplicate implementation effort, but (a) they encourage sharing ofalgorithms and code, and thus also cultivate increased collaborationand intellectual flow of ideas; (b) they foster the communication ofdetailed clarity of algorithms and scientific reproducibility; (c)they help "level the playing field" by providing state-of-the-artimplementations of foundational building blocks and recent methods totop-tier and small institutions alike; (d) they supply a teachingtool, not only by making it easy for students to experiment with thesupplied research methodologies. Furthermore, by providing multipleready-to-use systems, non-programmers will have access to modern,scalable implementations of text processing tools that will spreadknowledge and use of these techniques across fields, to the socialsciences, humanities, and bio-medical fields.For further information see the project web site at the URL:http://www.cs.umass.edu/~mccallum/nsf-mallet

自然语言处理、信息抽取、信息集成和其他文本处理解决方案是计算机科学的核心组成部分，也是解决日益严重的信息过载问题的关键工具。 信息过载问题不仅是个人问题，而且对企业生产力、国防以及越来越多的政府决策和透明度至关重要。最先进的自然语言处理越来越多地基于机器学习。 然而，这些方法可能是复杂的，并且这些系统所需的软件基础设施通常很难从头开始开发。 为了满足这一需求，我们创建了MALLET（MACHINE LEARNING FOR EQUIPAGE）和FACTORIE（Factorgraphs，Imperative，Extensible），这是一个运行在Java虚拟机上的开源软件工具包。 他们提供了许多现代最先进的机器学习方法，特别针对自然语言数据的特性进行了调整，同时也适用于许多其他离散的非语言任务。（1）扩大这些工具包对新数据和任务的适用性（具有更好的终端用户界面，用于标签、培训和诊断），（2）大大增强其研究支持能力（提供灵活指定模型结构的基础设施），以及（3）提高它们的可理解性和支持（提供新的文档、示例、在线社区支持）。该项目将对NLP和其他机器学习研究、教学以及合作研究活动产生直接的积极影响。 设计良好的工具包不仅可以帮助研究人员避免重复的实现工作，而且（a）它们鼓励算法和代码的共享，从而也培养了更多的合作和思想的智力流动;（B）它们促进了算法的详细清晰度和科学可重复性的交流;（c）它们通过提供最先进的基本构件和最新的方法来帮助“创造公平的竞争环境”，（d）它们提供了一种教学工具，不仅使学生容易试验所提供的研究方法。 此外，通过提供多个即用型系统，非程序员将能够使用现代的、可扩展的文本处理工具，这些工具将在社会科学、人文科学和生物医学领域传播这些技术的知识和使用。http://www.cs.umass.edu/~mccallum/nsf-mallet