Industry-University Co-operative Research Program: Acquisition and Use of Semantic Information for Natural Language Processing

产学合作研究项目：自然语言处理语义信息的获取和利用

• 批准号: 8501843
• 负责人: Ralph Grishman
• 金额: $ 17.99万
• 依托单位: New York University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1985
• 资助国家: 美国
• 起止时间: 1985-08-01 至 1988-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8501843&HistoricalAwards=false
• 关键词: Industry; University; Co; operative; Research

This project emphasizes the acquisition and use of semantic information, focusing on robust parsing of natural language inputs (in limited domains). Further, this research addresses the problems of portability, reliability, and habitability. It outlines the development of a set of interactive tools which provide, to both the linguist and the end-user, an accurate set of domain-specific co-occurrence patterns (subject-verb-object and host-modifier combinations allowed in the domain). This set of tools will partly automate the creation of the Domain Information Schema (DIS), namely, the domain-specific information required for processing a given domain. The same set of tools, provided with a different interface for the end-user, will be used to support interactive text entry. The tools will provide detailed feedback on where the system fails to process the text, due to missing or incompatible semantic co-occurrence patterns.

该项目强调语义信息的获取和使用，侧重于自然语言输入的鲁棒解析（在有限的领域）。此外，本研究还解决了可移植性、可靠性和可居住性等问题。它概述了一组交互式工具的开发，这些工具为语言学家和最终用户提供了一组精确的特定于领域的共现模式（领域中允许的主-动-宾和主-修饰语组合）。这组工具将部分地自动化域信息模式（DIS）的创建，即处理给定域所需的特定于域的信息。为最终用户提供不同界面的同一组工具将用于支持交互式文本输入。这些工具将提供详细的反馈，说明系统在哪些地方由于语义共现模式缺失或不兼容而无法处理文本。