CI-P: The ICSI Metaphor Annotated Corpus

这项CRI计划拨款使用概念隐喻理论来规划创建一个涵盖多种语言和语料库的在线隐喻资源。隐喻在人类语言和思维中无处不在，对语言表达和人类推理都至关重要。在日常生活、谈话、报纸以及包括科学、技术和文学写作在内的许多其他文本类型中，常规的和新颖的隐喻表达经常会遇到。该计划的目标是解决语料库选择和准备相关的问题，并在隐喻注释的迭代设计和实现中纳入社区广泛的评估反馈。在计划工作即将结束时，将有一个研讨会来评估方法，为语料库创建企业建立支持，并建立用户社区。这个语料库选择和隐喻注释工作也正在与正在进行的SGER合作项目（NSF IIS grant0843275）进行协调，该项目正在探索与可扩展隐喻建模和推理相关的问题。多层注释将包括相关词汇目标的概念关系和隐喻域、语言表达和框架注释以及在注释的目标文本中获得的句法关系。其结果将是设计一个全面的、注释丰富的隐喻语言资源。这项工作的结合，以及在语义分析和推理方面的并行努力，应该允许在几年内开发和测试能够进行比喻语言处理的大规模系统。

This CRI planning grant uses the theory of conceptual metaphor to planthe creation of an on-line metaphor resource that covers multipletypes of language and corpora. Metaphor is pervasive in human languageand thought and is crucial for both linguistic expression and humanreasoning. Conventionalized as well as novel metaphorical expressionscan be routinely encountered in everyday life conversations, innewspapers, and in many other text types including scientific,technical, and literary writing.The goal of this planning effort is to resolve issues related tocorpus selection and preparation, and to incorporate community wideevaluative feedback in the iterative design and implementation ofmetaphor annotations. Towards the end of the planning work, there willbe a workshop to evaluate the approach, to build support for thecorpus creation enterprise, and to establish a user community. Thiscorpus selection and metaphor annotation effort is also beingcoordinated with the companion SGER effort underway (NSF IIS grant0843275) that is exploring issues related to scalable metaphormodeling and inference.The multi-layer annotations will include conceptual relations andmetaphor domains, linguistic expressions and frame annotations of therelevant lexical targets, and syntactic relations that obtain in theannotated target text. The result will be the design of acomprehensive, richly annotated resource of metaphoric languge.Thecombination of this work, along with parallel efforts in semanticanalysis and inference should permit the development and testing oflarge-scale systems capable of figurative language processing within afew years.