Collaborative Research: Basic and Applied Research Leading to a Linguistic Probability Translator (LPT)

合作研究：基础和应用研究导致语言概率翻译器（LPT）

• 批准号: 9975360
• 负责人: David Budescu
• 金额: $ 18万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-09-15 至 2003-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9975360&HistoricalAwards=false
• 关键词: Collaborative; Research; Basic; Applied; Leading

This research aims to develop a Linguistic Probability Translator (LPT), initially as a useful research tool and subsequently for use as a decision aid within a communication community (defined as a collection of providers and receivers of risks estimates or forecasts). The LPT will improve the communication of subjective judgments and forecasts from experts to decision makers (DMs), and therefore also improve the resulting decisions. When fully realized, LPT will use quantitative methods based on empirical research to translate from one or many experts' individualized lexicons of uncertainty to a DM's individualized lexicon in a manner that best preserves the intended numerical and non-numerical meanings of the communication. If required, the LPT will translate the phrases into numerical probabilities or probability intervals, or from a numerical forecast into an appropriate phrase in a DM's lexicon. If there are multiple forecasts, LPT could translate each one into the DM's best phrase, or aggregate them and select a single best expression.The creation of LPT will require basic and applied research, as well as software development. The basic research will focus on issues of measurement and human information processing regarding the formation and communication of uncertain judgements. The software development will focus on the actual construction of LPT, and the applied work will examine the effectiveness and quality of the system. Overall, this research will advance basic theory regarding the measurement and communication of opinion, and will result in a computerized decision aid to improve decision making in the field.

本研究旨在开发一种语言概率翻译器（LPT），最初作为一种有用的研究工具，随后用作交流社区（定义为风险估计或预测的提供者和接受者的集合）中的决策辅助工具。LPT将改善专家与决策者（DMs）之间的主观判断和预测的沟通，因此也改善了最终的决策。当完全实现后，LPT将使用基于实证研究的定量方法，以最好地保留沟通中预期的数字和非数字含义的方式，将一个或多个专家的不确定性个性化词汇翻译成DM的个性化词汇。如果需要，LPT会将短语翻译成数值概率或概率间隔，或者将数值预测翻译成DM词典中的适当短语。如果有多个预测，LPT可以将每个预测翻译成DM的最佳短语，或者将它们汇总并选择一个最佳短语。LPT的创建将需要基础和应用研究，以及软件开发。基础研究将集中在不确定判断的形成和传播方面的测量和人类信息处理问题。软件开发将侧重于LPT的实际构建，应用工作将检验系统的有效性和质量。总体而言，本研究将推进关于意见测量和沟通的基础理论，并将导致计算机化决策辅助，以改善该领域的决策制定。