推理

According to the system view ofdegradation and decomposition put forward above,the paper proposes a newSplit Reasoning Simulation for qualitative model algorithm called SRSQM,which implements the decomposing,reasoning and solving of qualitative models.In another word,it combines variable cluster and causality graph to decompose acomplex qualitative model into many componentsand constraintsamong those components.Through the qualitative reasoning of the componentsand globally propagating the results,the overall solutions can be achieved.

根据复杂系统的递阶分解的系统观点，本文提出一种新的定性模型分解推理仿真算法，称作SRSQM，实现对定性模型的分解推理求解，即对于一个复杂的定性决策模型，应用变量聚类和因果图相结合的方法，将复杂的定性模型分解成若干个单元及其单元之间的约束，通过对各个单元的定性推理以及将单元推理的结果进行全局传播和过滤，得到整个系统的全局解。

The existential import is needed only when we make the following three kinds of inferences in categorical logic: conversion by limitation,the syllogisms with existential conclusions and two universal premises,and the inferences of contrariety an...

词项逻辑中涉及预设的地方仅为有限量换位推理，两个全称前提推出特称结论的三段论推理及反对关系推理和下反对关系推理所需预设均为：主项存在。从操作层面上看周延规则不是普遍适用的；从深层次来看，如考虑到预设，那么周延规则是普遍适用的。

Can be canceled or be invalid. Defeasible reasonings which have influence in many fields as an independent reasoning mode is supported by three theory to happen.

可辩驳推理作为一种独立的推理模式而在各个领域发生影响主要有以下三种理论的支持：其一，哲学的"语用学转向"；其二，人工智能研究的深入；其三，单一性法律推理向非单一性法律推理的转向。

First, 'linguistic turn' of philosophy; Second, Deepening studying in artificial intelligence; Third, monotonic legal reasoning turn to non-monotonic legal reasoning. Defeasible reasoning is widely used in legal reasoning.

可辩驳推理在法律推理中得到广泛的应用，根据法律推理的不同层次，法律中的可辩驳推理可以分为三个维度：推定的可辩驳性；过程的可辩驳性和理论的可辩驳性。

As one kind of compound reasoning, disjunctive reasoning has been full researched by the domestic researchers, but the related study on disjunctive reasoning under uncertainty is not so rich. Therefore, it is worthy of research.

析取推理作为一种复合推理，国内学者已有研究，但在不确定情景下，析取推理的表现以及影响推理的因素研究较少，值得对此进行探索。

Reasoning by analogy contain broad sense and narrow sense. The broad sense analogy indicates generally the inference based on genus, which equal to what we now say inference (contain deduce, induce, analogy). The narrow sense analogy only points several typical inference form based on same genus that the ancients discussed in Mohist Argument: pi, mou, yuan, tui, which mostly equal to what we now say analogy.

推类有广义和狭义之分：广义推类泛指在类的基础上进行的推理，相当于我们现在所说的推理；狭义推类仅指《墨辩》中论及的譬、侔、援、推等几种典型的、以类同为依据的推理类型，大致相当于当今所说的类比推理。

However, the support of stack-based function call is weak in Hoare logic.

此方法源于Hoare逻辑的程序推理以及耶鲁大学的认证汇编编程，它首先需要定义一个推理需要的目标机器和一组静态推理规则，并且证明推理规则的可靠性。

The"incomplete reasoning-verification"inference idea is used to form the synthetrical inference pattern with data-driven and object-driven. All of them increase the effectiveness of this algorithm greatly.

在推理策略上，运用了"不完全推理-验证"的推理思想，形成了数据驱动和目标驱动相统一的混合推理模式，从而大大提高了理解算法的效率。

In all, the results show that the information inference in discourse comprehension is inevitable, and it can happen on-line not only in working memory, but also in long-term memory.

本研究的结果表明，信息精确化推理在文本阅读中是必然发生的，而且不管进行需要精确化推理的句子与包含推理所需信息的句子是处于同一工作记忆范围内，还是处于长时记忆中，推理都是即时发生的。

Through analyzing a lot of current Ontology reasoners in detail, this paper concludes a general system structure for Ontology reasoners. After introducing three typical Ontology reasoners (Pellet, Racer, FaCT++), it proposes and implements a test plan for Ontology reasoners from the system, user and developer's point. The experiment results show that the test plan is feasible and effective. At last, the paper analyzes the disadvantages of current Ontology reasoner and gives some ideas on Ontology reasoners' future development.

通过对当前一些主流本体推理机详细的分析研究，得出本体推理机的一般系统结构，在介绍三个典型的推理机系统（Pellet， Racer， FaCT++）后，从系统功能，用户和开发者三个不同角度设计并实现一套比较不同本体推理机的测试方案，实验证明测试方案是可行有效的，最后总结当前本体推理机存在的一些问题和未来发展趋势。

I smile at your play with that little bit of a broken twig.

我微笑着看你玩那一段小小的树枝。

They may be produced for almost any shaft angle.

他们可能在制作中几乎任何一个轴角。