权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

CORG - Cognitive Reasoning

CORG-- 认知推理

基本信息

批准号：
388853480
负责人：
Professor Dr. Ulrich Furbach
金额：
--
依托单位：
Institut für Informatik
依托单位国家：
德国
项目类别：
Research Grants
财政年份：
2017
资助国家：
德国
起止时间：
2016-12-31 至 2020-12-31
项目状态：
已结题

来源：
https://gepris.dfg.de/gepris/projekt/388853480?language=en
关键词：
CORG Cognitive Reasoning

项目摘要

Cognitive computing addresses problems characterized by ambiguity and uncertainty, meaning that it is used to handle problems humans are confronted with in everyday life. When developing a cognitive computing system which is supposed to act human-like we cannot rely on automated theorem proving techniques alone, because humans performing commonsense reasoning do not obey the rules of classical logics. This causes humans to be susceptible to logical fallacies, but on the other hand to draw useful conclusions automated reasoning systems are incapable of. Humans naturally reason in the presence of incomplete and inconsistent knowledge and are able to reason in the presence of norms as well as conflicting norms. The versatility of human reasoning illustrates that any attempt to model the way humans perform commonsense reasoning has to use a combination of many different techniques.This project aims at the construction of a cognitive computing system by modeling aspects of human reasoning like emotions and human interactions. We will extend classical logical reasoning with non-monotonic reasoning like defeasible and normative logics in combination with machine learning. This will not only be carried out on a theoretical level. Different components for modeling the commonsense reasoning process will be developed and combined to a cognitive computing system which will be tested using benchmarks from commonsense reasoning.We want to address the following challenges:(C1) Finding Appropriate Logics for Cognitive Computing: There are essential differences between human and automated reasoning: Humans are able to reason with incomplete and inconsistent knowledge and naturally take background knowledge into account. Logics used for cognitive computing have to capture this versatility of human reasoning.(C2) Dealing with Large Background Knowledge Bases: Cognitive reasoning requires enormous amounts of background knowledge describing everyday experience humans use for reasoning. This background knowledge has to be constructed by combining appropriate sources. Further, the cognitive system must contain mechanisms to deal with the sheer size of this knowledge.(C3) Reasoning with Multiple Formats: Logical reasoning alone is not sufficient to model human reasoning. A cognitive system has to be able to handle natural language and deliberate about different conclusions which may be conflicting. For this, different reasoning techniques have to be combined in an infrastructure that ensures efficient cooperation.These challenges arise when constructing a cognitive computing system but have not been sufficiently addressed in the state of the art. We address them by combining reasoning mechanisms for logics suitable for cognitive computing, large amounts of background knowledge, and other techniques like machine learning. This will result in a cognitive reasoning system able to address problems which none of the techniques alone would have been able to address.

认知计算解决了以模糊性和不确定性为特征的问题，这意味着它被用来处理人类在日常生活中面临的问题。在开发认知计算系统时，我们不能仅仅依靠自动化的定理证明技术，因为执行常识推理的人不遵守经典逻辑的规则。这导致人类容易受到逻辑谬误的影响，但另一方面也会得出自动推理系统所不能得出的有用结论。人类自然而然地在不完整和不一致的知识存在下进行推理，并能够在规范和冲突的规范存在下进行推理。人类推理的多功能性表明，任何试图对人类进行常识推理的方式进行建模的尝试都必须使用许多不同技术的组合。该项目旨在通过对人类推理的各个方面进行建模，如情感和人类互动，来构建一个认知计算系统。我们将结合机器学习，用可废止逻辑和规范逻辑等非单调推理来扩展经典逻辑推理。这不仅仅是在理论层面上进行的。为常识推理过程建模的不同组件将被开发并组合到认知计算系统中，该系统将使用常识推理中的基准进行测试。我们希望解决以下挑战：(C1)为认知计算找到合适的逻辑：人类和自动推理之间有本质的区别：人类能够使用不完整和不一致的知识进行推理，并自然地考虑背景知识。用于认知计算的逻辑必须捕捉人类推理的这种多功能性。(C2)处理大型背景知识库：认知推理需要大量描述人类用于推理的日常经验的背景知识。这种背景知识必须通过组合适当的来源来构建。此外，认知系统必须包含处理这种知识的巨大规模的机制。(C3)多种形式的推理：仅有逻辑推理不足以模拟人类的推理。认知系统必须能够处理自然语言，并仔细考虑可能相互冲突的不同结论。为此，必须将不同的推理技术组合在一个基础设施中，以确保有效的合作。这些挑战在构建认知计算系统时出现，但在最新技术水平下尚未得到充分解决。我们通过结合适合认知计算的逻辑推理机制、大量背景知识和其他技术(如机器学习)来解决这些问题。这将导致认知推理系统能够解决仅靠任何一种技术都无法解决的问题。