原站英文机器转录 + 本地中文机器初译(未完成逐句听审) 原作者及许可见 attribution.txt;中文由九天新增,不代表原作者认可。 00:00:11.780 EN: So reasoning allows machines basically to support us in determining the clarity of the semantic representations and you have knowledge about all. 中文初译: 所以推理可以让机器 基本上支持我们 确定语义表达的清晰度 你对一切有了解 00:00:26.660 EN: Today we will talk a bit about description logics. Description logics are the foundation, the logical foundation of ontologies and then there is a reasoning algorithm which is used. 中文初译: 今天,我们将谈一些描述逻辑。描述逻辑是本体论的基础和逻辑基础,然后使用推理算法。 00:00:43.280 EN: We have learned with RDF and RDF schema the rule-based approach where you can apply certain rules to infer new information to reveal implicit information from RDF, RDF schema, vocabularies and simple ontologies with more complex one with all ontologies. 中文初译: 我们从“RDF”和“RDF”中学会了基于规则的方法, 你可以运用某些规则来推断新的信息以揭示来自“RDF”、“RDF”的暗中信息。 词汇和简单的本体论都包含所有本体论。 00:01:03.960 EN: This rule-based inferencing doesn't work anymore and there description logic reasoner use tableau algorithms and I want to show you a bit how that works, but we will not go too much into detail there just to give you a bit of an idea how reasoning with more complex ontologies work, but of course that's a topic which could fill its own lecture 中文初译: 这种基于规则的推论不再起作用了 并且描述逻辑理性 使用表情算法, 我想向大家展示一下它是如何工作的, 但是我们不会太详细到那里去 只是给你们一个想法 如何用更复杂的本体论理论来进行推理工作, 但当然,这是一个可以填满自己演讲内容的主题 00:01:32.080 EN: and it's a very active research field where you have hundreds or even thousands of researchers working on efficient reasoners and implementations there and that's particularly important when you have larger ontologies for very simple ones maybe we humans can discover inconsistencies, but once you have larger ontologies and more complex representation formalisms used the expressivity of OWL then it's not always clear or for humans clear 中文初译: 这是一个非常活跃的研究领域 在那里,有上百甚至数千名研究人员 在高效的理性和执行方面工作 这特别重要 当你拥有更庞大的理论论 来研究非常简单的理论时 00:02:07.760 EN: whether something is consistent or inconsistent and reasoners basically can assist us in finding such inconsistencies. 中文初译: 究竟是某种前后一贯或不一致的事物,理智者基本上能够帮助我们找到这种不一致之处。 00:02:17.700 EN: So that's the goal for today. Basically the semantics of OWL is based on description logics and they have a model theoretic formal semantics. 中文初译: 这就是今天的目标。基本上“OWL”的语义是基于描述逻辑, 并且它们有一个模型理论化的正规语义。 00:02:30.720 EN: So OWL DL for example corresponds to description logics showing D which you can see here OWL 2 corresponds to zroik D. 中文初译: 例如,“OWL” DL与描述逻辑相对应, 显示您在这里看到的 D “ OWL” 2 与 Zroik D 相对应。 00:02:48.520 EN: I don't know stand for certain features of these description logics. So and we will look at the simpler one showing D today in the lecture. 中文初译: 我不知道这些描述逻辑的某些特征。所以我们将看看今天演讲中演示D的更简单的逻辑。 00:03:00.240 EN: So description logics are a family of knowledge representation languages. They are usually fragments of first order logic. 中文初译: 所以描述逻辑是知识表达语言的组合。它们通常是一阶逻辑的零碎部分。 00:03:09.740 EN: First order logic also has the acronym FOL. They are in most cases decidable. So decidable fragments. Decidable means you can always answer certain questions about the logic whether something is true or false. 中文初译: 第一顺序逻辑也有首字母缩略语 FOL 。 它们在大多数情况下是可变的。 如此可变的碎片。 可变的意思是您总能回答某些关于逻辑的疑问, 无论什么是真实的还是虚假的 。 00:03:28.680 EN: That means decidable that you can in finite time decide certain questions. They are comparatively expressive. So description logics try to maximize this fragment of first order logic. 中文初译: 这意味着您可以在有限的时间内决定某些问题。 它们相对具有表达性。 所以描述逻辑试图尽量扩大第一顺序逻辑的这一片段 。 00:03:47.200 EN: First order logic overall is not always decidable. There are certain questions which you cannot answer in finite time. 中文初译: 第一顺序逻辑整体并非总能分辨。 有些问题您无法在有限的时间内回答 。 00:03:56.300 EN: So description logics try to always stay in this decidable fragment but at the same time be as expressive as possible. 中文初译: 所以描述逻辑试图永远 留在这个可分层的碎片中, 同时尽可能地表达。 00:04:04.940 EN: But of course we have to limit ourselves a bit to stay in this decidable fragment. They are originated from semantic networks. 中文初译: 但是我们当然得有所限制 才能留在这个可分层的碎片中。它们来自语义网络。 00:04:17.880 EN: They also have a new syntax which is a bit more intuitive. So because it doesn't use variables. So it's a variable free syntax. 中文初译: 他们也有一种新的语法, 更直观一些。 所以因为它没有使用变量。 所以它是一个可变自由语法。 00:04:31.200 EN: For example you write p is a sub class of q here. For example in this example instead of writing for all x for all variables x if p of x holds then q of x holds. 中文初译: 例如,您在此写入 p 是 q 的子类。 例如, 在此示例中, 如果 p of x 持有, 那么 q of x 持有, 则不对所有变量 x 都写入所有 x 。 00:04:48.560 EN: So that would be the first order logic notation and the description logic notation is just such a sub class notation. 中文初译: 因此,这将是第一个顺序逻辑符号,而描述逻辑符号只是这样一个子分类符号。 00:05:02.700 EN: It's a bit inspired also by subset. Subset or equal p subset or equal q because you can see classes as kind of sets of instances as well. 中文初译: 它也有点受子集的启发。 子集或相等子集或等q, 因为您也可以将类视为类似的例子。 00:05:17.040 EN: So the basic components of description logics are concept names. So we have atomic concepts like student book. So these are also called resources in the RDF world instances. 中文初译: 因此描述逻辑的基本组成部分就是概念名称。所以我们有像学生书这样的原子概念。在“RDF”世界实例中,这些也被称为资源。 00:05:40.540 EN: Then we have atomic concepts are classes in the RDF and RDF schema world. In description logics called concepts or concept names or atomic concepts. 中文初译: 然后我们有了原子概念,在“RDF”和“RDF”的理论世界中是等级。在描述学上,逻辑称为概念或概念名称或原子概念。 00:05:55.940 EN: Then we have role names and role names corresponds to what is called properties. Or predicates in the triples like born in or works for. 中文初译: 然后我们有角色名称和角色名称与所谓的属性相对应。或者在三进制中的上游,比如出生于其中或为之工作。 00:06:07.880 EN: And then we have individual names also called individuals or objects or in the RDF world instances or resources like Stephen and Mary for example. 中文初译: 然后我们还有个人的名字 也叫个人或物体 或者在“RDF”世界事件中 比如Stephen和Mary之类的资源 00:06:21.620 EN: So the set of concepts role and individual names is often denoted as signature or vocabulary of the knowledge base. 中文初译: 因此,一套概念作用和个别名称往往被指为知识库的签字或词汇。 00:06:33.860 EN: So a knowledge base, what is a knowledge base? A description logic knowledge base consists of two parts usually. 中文初译: 所以知识库,什么是知识库?描述逻辑知识库通常由两部分组成。 00:06:42.680 EN: One part is called t box and denoted with this tau for example and that's information about concepts. And then we have an a box denoted with a and that denotes information about individuals. 中文初译: 其中一个部分叫做t框,用这个图文来表示,比如说, 那就是关于概念的信息。然后我们有一个盒子,上面写着个人的信息。 00:06:58.280 EN: And for more expressive description logics there's also something called r box, r which contains information about roles. 中文初译: 对于更清晰的描述逻辑来说, 也有一些叫做 r 框的东西,r 包含关于角色的信息。 00:07:07.840 EN: In simpler description logics the roles are just part of the t box but in more expressive description logics you have more axioms also about roles and then you can call or separate that into an r box. 中文初译: 在更简单的描述逻辑中,角色只是t框的一部分,但在更清晰的描述逻辑中,你对角色也有更多的共性,然后你可以将角色调用或分离成一个 r框。 00:07:27.530 EN: So one of the simplest description logic is ALC and ALC stands for attribute language with complement. It's the most simplest propositionally closed description logic which subsumes propositional logic basically. 中文初译: 所以最简单的描述逻辑之一是 ALC和ALC 代表属性语言与补充。这是最简单、最封闭的假设描述逻辑,基本上包含假设逻辑。 00:07:50.550 EN: And complex or ALC concepts are inductively defined as follows. So every concept name is a concept so classes are concepts that's clear. 中文初译: 复杂概念或ALC概念的细微定义如下。所以每个概念的名称都是一个概念,因此类别是清晰的概念。 00:08:04.110 EN: Then two concepts are added the top concept and the button concept and they have the symbols top and the other one here bottom. 中文初译: 然后添加两个概念, 顶部概念和按钮概念, 它们有符号顶部, 另一则在底部。 00:08:12.970 EN: Yeah in our this corresponds to our thing and our nothing if you remember from two weeks ago yeah our thing is the class of subsuming everything and our nothing is the class corresponding to an empty set. 中文初译: 是的,在我们的这个对应 我们的东西和我们什么都没有 如果你记得两周前的事情。 是啊,我们的东西是 囊括一切的阶级 00:08:32.890 EN: So that's also relatively simple and then the third rule is if c and d are concepts and r is a role then the following are also concepts so we can build the negation of c the complement yeah we can build the conjunction of c and d denoted with this yeah this symbol here also the intersection or an end combination of c and d yeah so it basically builds 中文初译: 所以,这也相对简单。 第三个规则是,如果c和d是概念和r是一个角色 那么以下也是一些概念 所以我们可以建立对c的否定 00:09:07.470 EN: the intersection of the sets of instances of those two classes c and d. For example if you have the class of humans and the class of students yeah you can build the intersection or that doesn't make so much sense because it actually corresponds to the class of students you could also take I don't know the class of students and the class of employees and the intersection of that 中文初译: 比如,如果你有人类阶级和学生阶级... 你可以建造十字路口 或者说这没有意义 因为它实际上与学生阶级相对应 00:09:40.580 EN: would be the students who are working in addition to studying yeah. So that's the intersection and then we also have the union that's what you see here below also symbolized by this square union symbol and that's a disjunction a union or an or combination of the two concepts. 中文初译: 也就是除了学习是,还在工作的学生。这就是十字路口。然后我们还有工会, 这就是下面你看到的, 也代表着这个正方形的工会符号, 是一个分离的工会, 或者一个或两个概念的结合。 00:10:05.240 EN: So you could build the class here also of or the concept and uniting for example the students and the employees and then it would contain all students and all employees yeah so that's the disjunction where a member of the new concept needs to be only a member of one of the two constituents in that case c or d. 中文初译: 因此,你可以在这里建起一个班级,比如学生和雇员的概念和联合, 然后它会包含所有的学生和所有雇员。 是的,所以这是一个不相干的地方, 新概念的成员在 c 或 d 的情况下, 只需要成为其中两个成员之一的成员。 00:10:35.860 EN: Then a third possibility to build new concepts is an existential restriction. So we it's denoted with it exists r and c so that means it exists an instance or it builds basically a set of instances. 中文初译: 然后,建立新概念的第三种可能性是存在限制。所以我们用它来表示它的存在 r和c 所以它意味着它存在一个实例 或者它基本上构建了一套实例。 00:11:02.810 EN: You remember we had in the our lecture the um some values some value of restriction and that's such an existential restriction. 中文初译: 你记得我们在演讲中说过 嗯 一些价值 某些限制的价值 这就是存在的限制 00:11:17.650 EN: So it builds the set of instances which have where the instance has the property r and the value of this property belongs to the class c yeah. 中文初译: 所以它构建了一组实例, 这些实例中 属性 R 和这些财产的价值属于 c 类。 00:11:31.030 EN: So it basically selects a set of instances and defines this as being a new concept. It selects all instances which have the property r as one property with a value and the value belongs to c. 中文初译: 因此它基本上选择了一组实例, 并将其定义为一个新概念。 它选择了所有将属性 r 作为具有一个价值和数值属于 c 的同一属性的所有实例。 00:11:50.050 EN: What happens c for example can be also the top concept then it basically means the value doesn't matter then we select all instances which have a property r yeah. 中文初译: 比如说,发生的事情 c 也可以是顶级概念 然后它基本上意味着价值并不重要 然后我们选择所有有属性 r 的事例 00:12:03.030 EN: Doesn't matter what the object of these triples refers to if we have a concrete c and not the top concept then the value of r must be the object of the or the value of this property must belong to this class c. 中文初译: 这些三重物的对象指什么无关紧要 如果我们有一个具体c而不是顶级概念 那么r的价值必须是该财产的标的 或者价值必须属于这个 c类。 00:12:23.050 EN: And the selection of all those instances then is the set of instances belonging to our new concept. And then we have a last construction of new concepts that's for all r. 中文初译: 选择所有这些例子 属于我们的新概念。然后我们有最后一个新概念的构思, 这个新概念是适合所有r。 00:12:44.190 EN: The values of r must belong to c so that's a value restriction and we had an owl this all value from restriction so that means all values of instances which have the property r must belong to the class c. 中文初译: r的值必须属于 c,所以这是一个价值限制 我们有一个猫头鹰 这一切的价值 从限制 所以这意味着所有具有 r 属性的情况的所有值 必须属于 c 类。 00:13:02.490 EN: So we built a new concept by selecting basically all these individuals which have the property r and where the values of this property r belong to c. 中文初译: 因此,我们建立了一个新概念, 选择了所有这些基本上具有产权 r 的个人, 而这些产权的值属于c。 00:13:16.370 EN: So an example for an extensional restriction for example would be the set of parents yeah we could say r is is parent off and then c could be human so we could that would be actually the set of children if we select all instances which or the the set of parents 中文初译: 举例来说,扩展限制的一个例子 就是一群父母... ...我们可以说r是父母的离家,c可能是人。c可能是人,所以如果我们选择了所有哪种或那组父母的情况,那实际上就是一组孩子。 00:13:49.110 EN: if you have the property is parent off and human so that would for example we could also have animal parents right but if we restrict this is parent off property to only those where the values are belong to the class humans then we basically define the set or the concept of fathers and mothers. 中文初译: 如果你的财产是亲生父母和人,我们也可以有动物父母的权利,但如果我们把这种财产的亲生父母仅限于那些价值属于人类阶级的人,那么我们就基本上界定了父亲和母亲的一套概念。 00:14:17.770 EN: So you can find some more examples for these so this is basically the definition of the ALC description logics very simple basically this inductive definition of course you can apply once you apply it these rules you can construct more complex concepts by iteratively applying more of these conditions here. 中文初译: 因此,你可以找到更多这方面的例子。 基本上,这是ALC描述逻辑的定义, 简单得多, 基本来说,你一旦应用了这个启发性定义, 你就可以应用这些规则, 通过在这里反复应用更多的这些条件来构建更复杂的概念。 00:14:43.230 EN: So here are some examples for example we have the concept person and then we have an intersection with a restriction on the property has child and then the top concept here so these are actually all persons with children yeah animals would be included also animals could have the property has child right but since we have the intersection with 中文初译: 举例来说,这里有一些例子 例如我们有概念人, 然后我们有一个交叉点 与对财产的限制 有孩子, 00:15:12.110 EN: the concept person this would build us a class basically of parents. So second example here we have animal intersection with for all eat eats must belong to the class vegetable so let's assume yeah we have such a concept animal and the property eats and then you have you can apply the property multiple times 中文初译: 这个概念人,这个概念人 将构建我们一个基本上由父母组成的阶级。第二个例子,这里我们有动物与动物交叉的 供所有食用的食物必须属于菜类 所以让我们假设我们有这样的概念动物 和财产吃,然后你可以多次使用这些财产 00:15:42.610 EN: yeah animals can eat leaves they can eat grass they can eat other animals they can eat meat they can eat insects and here we select those instances those individuals which have the property eat but which eat only vegetables and then we build the conjunction with the animal concept so these are the vegetarians 中文初译: {\fn华文楷体\fs16\1cHE0E0E0}这里我们选择了那些有财产但只吃蔬菜的人 {\fn华文楷体\fs16\1cHE0E0E0}然后我们把动物的概念结合起来 {\fn华文楷体\fs16\1cHE0E0E0}所以这些是素食者 00:16:16.310 EN: under the animals yeah another example is professor union student so this basically builds the concept of all we could say university members although maybe also employees so we would have to add another union if we also want to capture employees but this would be a concept of professors and students or students and the last example here also again 中文初译: 另一个例子是教授工会学生, 因此这基本上构建了所有我们可以说的 大学成员的概念 虽然可能还有雇员。 所以我们不得不加入另一个工会 如果我们也想抓捕员工 00:16:49.550 EN: a bit more complex so we have person and then the intersection or conjunction with for all properties born in the value must should must not belong to a european country yeah so this basically selects all individuals which have the property born in and where all values of this property born 中文初译: 更复杂一点,所以我们有个人, 然后交叉或连接 与所有产于价值中的所有财产的交汇或结合 必须不应该属于一个欧洲国家。 是的,所以这基本上选择了所有拥有生于该财产的个人, 以及这些财产的全部价值诞生的地方 00:17:18.410 EN: in point to an instance of which do not belong to a european to europe to the concept european country so this basically selects persons which are not born in a country in europe yeah and you see here the expression is already a bit more complex we actually have three combinations so we have the negation of european country here then we have the all values restrictions there and then we have 中文初译: 指出一个不属于欧洲的欧洲人到欧元的欧洲人与欧洲国家的概念。 因此,这基本上选择了不是出生在欧洲国家的人。 是的,这里你可以看到这个表达方式已经有点复杂了 我们实际上有三种组合。 所以我们在这里否定了欧洲人的国家 然后我们得到了所有的价值限制。 00:17:53.530 EN: a conjunction between person and the second concept definition here yeah so that illustrates how we can build concepts basically new classes out of existing classes or out of selecting instances based on properties in our knowledge base so what about the t-box um the t-box also terminology box consists of a finite set of terminological axioms 中文初译: 人与第二个概念定义之间的结合 来说明我们如何从现有类别中或者根据我们知识库中的属性选择实例, 从现有的类别中构建基本新类别的概念。 那么关于t-box um, t-box 也包含一系列有限的术语轴 00:18:26.450 EN: terminological axioms are basically general concept inclusion axioms for example or a given concept c and d these general concept inclusion axioms are denoted in this way here yeah c is a subclass of d and equality and basically in owl this would be subclass of yeah 中文初译: 术语的正exions基本上属于一般概念,包括轴轴(例如)或特定的概念c和d)。 这些一般性概念包含轴(例如)c和d。 这里用这种方式表示这些总概念包含轴(axions)。 eoh c 是一个d和平等的子类。 基本上在猫头鹰中,这将会是 i 的子类。 00:18:54.520 EN: c subclass of d and the concept equality can be expressed with this equal sign with three dashes here c equal d and this corresponds to the owl property equivalent class yeah so in all we can use the equivalent class property to indicate equality but in fact this is just syntactic sugar and this equality is just 中文初译: 与猫头鹰财产等同等级的等值等级是相同的。 因此,我们可以使用等等等类财产来表示平等,但事实上这只是混糖,这种平等只是 00:19:23.540 EN: an abbreviation for c is a subclass of d and d is a subclass of c so you could also write these two instead of the equality equality just allows you to write things down a bit more concise and even in owl this shortcut equivalent class is introduced for that purpose to make it more concise so what is the a-box um the a-box is our a finite set of assertional axioms 中文初译: C的缩略语是d和d的子类,d是c的子类,所以你也可以写这两个,而不是平等, 仅仅允许你写下来更简一些,甚至用猫头鹰写下来, 这个快捷的等效类被引入了这个目的, 以便让它更加简洁, 所以什么是 a-box um, a-box 是我们一套有限的直言轴 00:19:59.460 EN: of the following forms so we have basically two things we have concept assertions for example here c a belongs to c yeah this is a concept assertion this basically means a is of type c a the individual a belongs to the class c this is a concept assertion in rdf you would just write a rdf type c yeah and then we have a so-called 中文初译: 我们基本上有两样东西, 概念主张,例如这里 c a 属于c yea,这是一个概念主张, 基本意思是c 型个人属于C类。 这是在Rdf中的概念主张, 你会写一个rdf 类型c ye,然后我们有一个所谓的 00:20:28.320 EN: role assertions and there we write the role and then the two individuals which are connected by the role here a and b this is a role assertion and that means a um yeah has the property r with the value b yeah and rdf we would write a r b that would be the triple yeah so you can see ontologies on the one hand from the triple perspective 中文初译: 我们写角色声明,然后我们写这个角色 我们写了这两个人 与角色有关 这是角色声明 这意味着um yeah有财产r 价值b yeah和rdf 00:20:57.180 EN: everything is subject predicate object triples but you can also see them from the logical perspective and then there is this this notation this description logic notation here and the a-box basically contains of these kind of assertions concept assertions so individuals belonging to classes to concepts or role assertions connecting two individuals with each other so a formal definition of the model theoretic semantics of alc is 中文初译: 所有物体都是所有对象的谓词物三倍, 但你也可以从逻辑角度看到它们。 然后这个符号 说明这里描述的逻辑符号 和一个盒子基本上包含这类断言的概念主张 因此属于类别的个人 与概念或角色指称有关 00:21:30.980 EN: given by means of an interpretation yeah we had this already in rdf and rdf schema so an interpretation consists of two things um non-empty domain um symbolized by this delta and then a mapping which maps every individual a to a domain element um a in the interpretation of a from the domain every concept name uh capital a 中文初译: 通过解释的方式,我们已经有了这个在 rdf和rdf schema中。 因此,一个解释由两件事组成 um um - none-creat 域域域 um um um 00:22:04.880 EN: to a subset of domain elements yeah so the concepts are interpreted as subsets so basically sets of domain elements yeah sets of individuals and instances and every role to a set of pairs of domain elements uh so a role basically is a subset of the cartesian product of our domain 中文初译: 域元素的子集。 是的,所以这些概念被解释为子集。 基本上就是一组域元素的子集。 一群个人和实例,以及一系列域元素的每组角色。 呃,一个角色基本上是我们域的笛卡尔产品的一个子集。 00:22:38.770 EN: with itself yeah so basically a role is a set of tuples where the components of the tuples are belonging to our domain delta so that's then an interpretation an interpretation you can imagine or picture yourself like a concrete um concrete resources rdf resources the other domain for example can be a set of rdf resources 中文初译: 本身是的,所以基本上一个角色 是一套图普尔图普尔的成分 属于我们的域三角洲 所以这就是一种解释 你可以想象或想像自己 就像一个混凝土的具体资源 00:23:10.720 EN: and then we have a mapping of our logical concepts and individuals and roles to properties to classes um and to instances in our concrete um knowledge base so the interpretation is extended to complex concepts by um the following so the interpretation of the top concept is the domain 中文初译: 然后我们绘制了我们的逻辑概念 以及个人和作用的地图 并将其与属性的属性联系起来 分为平级和具体知识库中的实例 这样解释就扩大到复杂的概念 00:23:41.380 EN: itself yeah the domain basically is the the set of all uh things in our knowledge base like our thing for example in the owl terminology the bottom concept is the empty set here the negation of a concept or the interpretation of the negation of a concept is we take the domain and subtract and remove from the domain basically the concept c 中文初译: (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声(掌声) (掌声) (掌声) (掌声 (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声(掌声) (掌声) (掌) (掌声) (掌声) (掌声(掌声) (掌声) (掌声) (掌声) ( 00:24:15.520 EN: yeah so we remove all things which belong to c from our domain and that's the complement yeah then we have um here this um conjunction between c and d and the interpretation of that is that we built um the intersection of the two sets uh c um the interpretation 中文初译: (笑声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) (掌声) 00:24:47.100 EN: of c with the interpretation of d yeah so basically we just make the sets of instances belonging to the interpretation of c and intersect that with the uh set of instances belonging to the interpretation of d the same we do with the disjunction here the interpretation of the disjunction is the union of the interpretation of uh c 中文初译: 用对 d ye的诠释来解释 c 所以基本上我们只是把这几组情况 归为对 c 和 交叉的诠释 00:25:16.460 EN: and d yeah so we basically um the logical axioms what we do here we in the interpretation um allows us to to see the logical axioms in the light of set theory on the right hand side here you see only sets and set operations right on the left hand side of these equations we have a logical axioms on the right hand side we have just 中文初译: 和d是的,所以我们基本上,嗯,逻辑的轴法 我们在这里做的逻辑轴法 我们在这里的解释中,um 使我们能够看到逻辑轴法 从右手侧的定置理论来看 00:25:45.250 EN: set operations union um this um intersection the minus the set difference here yeah that's the uh what uh how we can apply these interpretations so let's look at the last two examples here we have the existential restriction how do we interpret that as a set so we basically select select all x in our domain all individuals 中文初译: 设置操作联盟, um 这个 um um 将设定的差值与此处的差数 相交叉 yeah, that's the uh 我们如何运用这些解释 所以让我们看看最后两个例子 00:26:14.770 EN: or all concepts in our domain for which there exists a y in the class c in the set which belongs to the set of the interpretation of c such that x y belong to the interpretation of r yeah that the tuple x y belongs to r yeah so this basically means we have a triple x r y where y belongs to the class c and we select all these x's 中文初译: 或所有概念 我们的域域中存在一个y 在一组C类中的 c类中有y 存在的所有概念 在一组c 的解释中,该组属于 c 解释组的 c , 因此x y 属于 r y 解释 r 的 r ye y 是 r y 的 r ye 。 这基本上意味着我们有一个 3 x r y 的 y y 属于 c 类, 我们选择所有这些 x 00:26:50.150 EN: and that's the new set a new set of individuals which is defined basically by this kind of axiom here by this existential restriction and then similarly we have the um these all values restriction here how do we interpret that again we select all um we we select all x in our domain delta where the following holds yeah for all x y 中文初译: 这就是新的一组新个人, 基本上由这个存在限制所定义的这种轴轴来定义。 然后类似地,我们这里有这些所有价值的限制 我们如何解释,我们再次选择 我们选择了我们域域三角洲中的所有的 um 我们选择了全部的 x, 在三角洲里,以下保留了... 00:27:21.390 EN: so for all um triples x r y or all tuples x y which belong to the role interpretation r r i here y must be belong to the class c yeah so in the existential restriction up here just one needed to exist yeah and that was sufficient for our x to belong to this class for our 中文初译: 所以对于所有 um 3x r y 或所有 Tuples x y y y 属于角色解释 r r i 这里 y 的角色解释 r r i y 的, y 必须属于 c 类 c 。所以在这里的存在限制中, 00:27:53.770 EN: all values restrictions all properties or all values of this property r must have values which belong to c so we again also here select a set of things which fulfill basically this restriction and we construct a new class in that way so interpretation is 中文初译: 此属性 r 的所有值限制所有属性或所有值。 此属性 r 的所有属性或所有值必须具有 c 所属的值, 因此我们在此再次选择一组基本满足此限制的事物, 我们以这种方式构建一个新的类别, 这样解释就可以了 00:28:22.250 EN: extended to axioms by the following rules the interpretation i fulfills c subclass of d if c if the interpretation of c is a subset of the interpretation of d yeah the interpretation i fulfills c equal d if uh the sets are equal yeah if c is exactly the same set as d an interpretation i 中文初译: (c) 如果对 c的解释是 d 的诠释的一个子集, i 的解释i 满足 c equal d 如果两套合等, uh 等于 uh;如果 c 与 d 解释一完全相同, i 00:28:58.950 EN: holds for c of a if a is an element of um the interpretation of c yeah a belongs basically to the concept um c and an interpretation um or r a b holds for an interpretation i if um the interpretation of a and b the tuple um is an element 中文初译: c , a 基本上属于um c 概念,而 um 或 r a b 则属于 i 解释, i , um 对 a 和 b 的诠释是一个要素。 00:29:31.060 EN: of the interpretation of r so again we basically trace back the um interpretations to set um set operations yeah so we can validate uh certain uh compare certain sets here check whether certain something is an element of a set and based on this checks um set theoretical checks we can determine 中文初译: 我们基本上可以追溯到对 r 解释的解释 来设置 um set 操作 是的 所以我们可以验证 呃 某些... 00:29:59.340 EN: if i is an interpretation for certain axioms so let's look at some example here so we have the following t box um man equals to um not woman um and this junction conjunction with person yeah so we basically select all um or we build here the disjunction between the negation of woman and person 中文初译: 如果我是某某定理的诠释 所以,让我们看看这里的例子 所以我们有以下的t盒子... um man 等于um not woman... 00:30:34.040 EN: and define that to be man so we define woman to be a sub-concept of person we define mother to be equivalent to a woman which is um a conjunction with the restriction um on the property has child with the value top concept 中文初译: 并定义作为男性,这样我们定义妇女为子概念 我们定义母亲等同于女性 与对财产的限制 00:31:02.080 EN: so some some value everything which has the property has child and that's intersected with the concept woman yeah so this is our t box so these are axioms which talk about concepts and um roles here and then we look at our a box the axioms so we have stephen belongs to the concept man monica belongs to the concept not 中文初译: 所以有些东西的价值是 拥有财产的一切都有孩子 而它和女人的概念交织在一起 是的,所以这是我们的t框。 所以我们的t框,这些是xixoms 讨论概念和um角色在这里, 然后我们看看我们的盒子 概念和um角色在这里, 00:31:34.660 EN: man so complement of man jessica belongs to the concept of woman and then we have um the role has child um on the role axiom and stephen is connected via has child to jessica yeah this is our a box here that's the way how we can build knowledge bases in alc in the simple description logic yeah so here we have this equivalent 中文初译: 男人的Jessica补充男子, 所以男人的Jessica 属于女人的概念 然后我们有了... 我们的角色有孩子... 00:32:06.700 EN: with the owl concepts yeah so the top i i mentioned that already top concept is all things button concept is all nothing the negation um corresponds to our complement of the union or this um uh conjunction corresponds to our union of the disjunction corresponds to 中文初译: 与猫头鹰的概念 是的,所以上面我提到 已经是顶级的概念 所有事物按钮概念 都不是否定的um 00:32:36.820 EN: our intersection of and the extensional restrictions to our sum values from and the um all values from restriction to all all values from so basically we can um translate this description logic in owl or vice versa we can translate owl into this description logic 中文初译: 我们的总值的交叉点和延伸限制 以及我们总值的连接点和扩展限制, 以及从限制到所有价值的所有值, 从如此基本上 我们可以将猫头鹰的描述逻辑 或反之亦然 我们可以把猫头鹰的描述逻辑 00:33:07.660 EN: and um the description logic also into um owl and um and rdf triples in the end yeah and that's a great uh great thing that we can can do that um because um basically this rdf triples that's something like a data um a database basically we have here a way to connect um a data representation with a logic yeah so then um there is another description logic alc plus or alc or alc 中文初译: 嗯,嗯,描述逻辑 也编成小猫头鹰和 额及大发三重 最后最后的3倍 是的,这是一个伟大的uh很好的事情 我们可以做到这一点,嗯 因为嗯,基本上这个大发三重 这就像一个数据 00:33:43.860 EN: plus inverse roles yeah and that's called alc i um and the i stands for inverse roles yeah so basically the names of the description logics are acronyms uh related to the features a role can be a role name r or an inverse role r minus semantics of inverse roles is defined by r minus the interpretation of r minus is all y x where x y belongs 中文初译: 加上反面角色 也就是所谓的alc i um 和 i 代表反面角色 yeah, 所以基本上描述逻辑的名称是缩写 uh 与一个角色的特征有关 00:34:14.000 EN: to the interpretation of r yeah so we basically built the opposite and the owl construct for that is inverse of yeah so um i guess you can all come up with some examples of inverse roles for example has parent and has child would be inverse yeah somebody who has a parent this person who is a parent has this somebody as a child yeah 中文初译: 对 R 的诠释是的,所以我们基本 建造了相反的,猫头鹰的构造 是反的。 所以,嗯,我想大家可以想出一些反作用的例子 比如说,有父母和有孩子 00:34:44.180 EN: so that's a typical example of an inverse role and you can then go in this different directions so then alc plus role hierarchy that's another feature which we can add to this description logic and the name is then a lch h for hierarchy yeah so and then if you have roles r and s and role inclusion axiom is denoted by r sub role 中文初译: 因此,这是一个典型的反作用的例子, 然后你可以沿着不同的方向走, 然后变法加上角色等级, 这是另一个特性 我们可以添加到这个描述逻辑中, 名称是等级的 lch h。 是的, 所以如果您有角色 r 和 s , 并且角色包含 Axiom, 则用 r 子角色来表示 00:35:16.500 EN: of s you see here it's the same symbol as for concepts for sub concepts or sub classes yeah so the symbol is here overloaded so don't be confused you have to be very careful to understand here in that case we always use for role small letters yeah and for concepts capital letters and then you basically can see when the symbol is applied to 中文初译: 这里可以看到它与子概念或子类的概念相同, 它和子概念或子类的概念相同。 是的, 所以符号在这里超负荷过载, 所以不要混淆了。 在此您必须非常小心地理解 00:35:46.530 EN: small letters it's a role inclusion axiom when it's applied to capital letters it's a class or concept um subsumption axiom so and then we also have the equivalent r equal s and it's an abbreviation for r is a sub role of s and s is a sub role of r do you remember we discussed this a bit in the whole lecture what could be sub roles so 中文初译: 小字母,它是一个角色包含轴法 当它应用到大写字母时 它是一个类或概念 um um susususimption axiom 所以,然后我们也有等同的r equal s 00:36:14.610 EN: for example being married to could be a sub role of being knowing someone right if you are married to someone then you definitely know this person as well so that's why being married to could be modeled as a sub role of b of knowing another person or being married to someone could also be a sub role of being related yeah we could have the role being related 中文初译: 比如说,结婚可能是了解某人的权利的一个次要角色, 如果你与某人结婚,那么你肯定也认识这个人。 所以为什么结婚可以被模拟为B了解另一个人或与某人结婚的次级角色 也可以成为关联的次要角色。 00:36:47.690 EN: with someone else and then we have different types of sub roles which could be being married to being child of and so on yeah as sub roles of being a relative of yeah there are different types of relationship family relationships and these could be modeled as sub roles and that's um then a role hierarchy and in owl um the owl construct 中文初译: 和别人在一起,然后我们有不同种类的子角色 可以结婚生孩子, 因此,当作为相亲的子子角色时, 有着不同类型的关系家庭关系, 这些关系可以建模为子角色, 也就是... 一个角色等级和猫头鹰的猫头鹰 00:37:17.190 EN: would be rdfs sub property of um you should remember that and basically an interpretation i entails r is a sub role of s if the role interpretation of r is a subset of the role interpretation of s what was the role interpretation it was the set of tuples right the set of tuples of individuals which are connected via this role and 中文初译: 你该记住 基本上解释i意味着r是s的子作用 如果 r 的作用解释是解释s的角色解释的一个子项 00:37:47.730 EN: if this set of tuples is a subset of the set of tuples of s then we have this role inclusion so it's always important to picture oneself this set theoretical foundation yeah roles properties are basically sets of tuples and the interpretation um then or you can build then subsets of of these kind of tuples and determine 中文初译: 如果这组图普尔是一组图普尔的子集, 那么我们就有了这个角色的包含, 所以让自己想象这个设置的理论基础总是十分重要的。 任务属性基本上就是图普和解释的组合,嗯, 或者你可以再建立这些图普尔的子集,然后确定 00:38:19.350 EN: if a role um is a sub role of another one for example it doesn't have to be always explicitly stated for example you can have two roles you have the role is married to and the role is related to right these can be two different roles and then a reasoner can actually discover if the um sets of these tuples of the roles are subsets that is married to is actually a sub role of um is related to yeah so you don't have 中文初译: 比如说,如果一个Um是另一个角色的一个子角色,那么它就不必总是明确说明,例如,你可以有两种角色,你的角色是已婚的,而角色与权利有关。 这两种角色可以是两个不同的角色,然后一位理智者可以实际发现,这些角色的图例中的那组是子子子,而结了婚的这个子角色实际上是um的一个子角色。 00:38:52.990 EN: to have it explicitly stated but the reasoner could figure that out that if in all cases people who are married with each other are also related with each other then a reasoner will discover that such a sub role relationship holds or the sub property of relationship next feature which we could add is role transitivity yeah so we add to alc 中文初译: 明确声明,但理智者可以发现,如果在所有情况下,彼此结婚的人也相互联系,那么一个理性者就会发现,这种次级角色关系或关系的子属性具有这种关系的下一个特征,我们可以加上的是角色过渡性。 00:39:22.190 EN: um the transitivity and this is now um now becomes the symbol s yeah so because the acronym would become otherwise too long um it just gets the symbol as alc plus transitivity is uh is the same um as s the description logic s so for a role r a transitive axiom is denoted by trans r so we basically determine in a role to be 中文初译: 即... ...mum... ...mum... 00:39:59.610 EN: transitive with this um the owl construct is our transitive property we have discussed that two two weeks ago and the interpretation of i entails trans r if for all x y in r and y z is also in r follows that x z is in r right so that's exactly the transitivity yeah if x y is uh connected x and y is connected with or belongs to the role 中文初译: 猫头鹰的构造是 我们讨论过的暂时性财产 我们两周前已经讨论过 i 的诠释意味着转r 如果对于所有 x y y in R和 y z 来说 00:40:34.230 EN: and y z then also x z belongs to the role if that holds then r is transitive or if r is transitive then this rule must hold and then we can also infer more tuples to belong to r basically based on the transitivity for example being related to someone is also transitive right if a is related to b and b is related to c then also a is related to c 中文初译: y z 和 y z 然后xxxz 也属于该角色,如果这样持有r 是过渡性的,或者如果r是过渡的,那么这一规则必须维持,然后我们还可以推断更多的图普尔属于r,基本上基于中转性,例如与某人有关联,如果与b和b有关联,则也属于过渡权,如果与c有关联,那么也与c有关联。 00:41:06.350 EN: so that's a transitive role and we could then discover new relationships between relatives um which we by just following the path of being related with each other so next feature is role functionality this is the description logic alcf so for a role r a functionality axiom is denoted by func r so r is functional and interpretation i entails um func r 中文初译: 所以,这是一个过渡性的角色, 然后我们就可以发现 亲属之间的新关系 嗯,我们只是沿着 彼此联系的路径 我们只是走 路径 彼此的关系 所以下一个特征是角色功能 这是描述逻辑 alcf 00:41:43.950 EN: if the following holds so we have x y belongs to the role interpretation and x z belongs to the role interpretation then y and z are equal yeah that means a role can only have one value or property can have only one value if it has different values those different values are actually referring to the same entity real world entity yeah in owl this was called owl functional property 中文初译: 如果以下的持有,我们拥有x Y 属于角色解释,x z 属于角色解释,那么y和z是等同的 ye 和 z 是等的。这意味着一个角色只能有一个价值,或者如果具有不同的价值,这些不同的价值实际上指同一个实体真实世界实体,那么,x y y 和 z 属于角色解释,y 和 z是等的。这意味着一个角色只能有一个价值,或财产只能有一个价值,如果它有不同的价值,则这些不同的价值实际上是指同一个实体真实世界实体。 猫头鹰中这个叫做猫头鹰功能属性 00:42:16.710 EN: so that's the description logic alcf so then we have a simple versus complex roles so if you have a role hierarchy r and then with the following symbol here we denote its reflexive and transitive closure a role r is complex with regard to r 中文初译: 因此,这是描述逻辑的逻辑 alcf 所以,我们有一个简单 和复杂的角色 所以如果你有一个角色等级制度r 00:42:48.160 EN: if there exists a role s such that the transitive closure of s is an element of r as an element of the role hierarchy and s is a sub role of r otherwise the role r is simple so basically a role is complex if there are other roles if there are sub roles of r so 中文初译: 如果存在这样一种作用,即暂时性关闭是r的一个要素,作为角色等级的一个要素,而这是r的子作用,或r的子作用,否则,r的作用就简单了,因此,如果存在其他角色,如果有r的次级作用,则基本上是一个复杂的角色。 00:43:22.360 EN: that r could be constructed after as a transitive closure of um of the sub role yeah otherwise we call a role simple if that doesn't exist so let's look at some example so we have a role hierarchy u is the sub role of r and r is a sub role of s and s is a sub role of t and q is a sub role of t and then we have the transit 中文初译: r 的子角色,r 的子角色,s 是 t 的子角色,q 是 t 的子角色,然后我们有 00:43:57.520 EN: and um r the role r is a transitive here in that case which ones are complex in that case r s and t are complex yeah r is complex because this has a sub role u s is complex because it has a sub role r and uh t is complex also because it has a sub role s and q here yeah 中文初译: r 和 r r 的作用在此具有过渡性,如果在这种情况下, r s 和 t 是复杂的, r 和 t是复杂的。 r 之所以复杂, 因为它有一个子作用 u s 是复杂的, 因为它有一个子角色 r 和 uh t 也是复杂的, 因为它在这里有一个子角色 00:44:29.230 EN: while u and q are simple because u doesn't have a sub role and q also doesn't have a sub role right so now another extension of our alc and that's um unqualified number restrictions uh so this is called a lcn number restrictions so for a simple role r and the natural number n um 中文初译: i和q是简单的,因为u没有子角色 而q也没有子角色 所以现在又延长了我们的 ac的扩展 00:45:04.410 EN: we can i have these number restrictions larger or equal nr smaller or equal nr or equal nr and these are concepts which are um defined which semantics is defined as follows so the interpretation of larger or equal nr the interpretation of that selects all x in our domain 中文初译: 我们可以有这些数量限制大于或等于 nr 小于或等于 nr 或相等 nr 或相等 nr, 这些概念是缩略定义的, 其语义定义如下, 这样对大或等 nr 的解释, 也就是对我们域中选择全部 x 的词义解释 00:45:37.440 EN: where the number of y's or so basically here this um this means counting yeah counting the set of um of this um uh of elements which are defined here so we um select all um individuals of our domain y where xy belongs to our role interpretation 中文初译: 这里的 y 或 y 的 数 。 这个 um 表示计算 。 计算 。 计算 。 00:46:07.840 EN: and if the number of these is larger or equal to n then we select the x to belong to this restriction yeah so that basically means we select all individuals which have the property r more equal or more than n times so for example um parents of um 中文初译: 如果这些数量大于或等于 n,那么我们选择属于此限制的 x 。 。 。 。 。 。 00:46:39.050 EN: of uh three or with three or more kids yeah we could define um this as um and then we could for example in that case r is um has um as children or has child has child would be r and n would be three so we could say larger or equal three has child what would this select this would select all individuals which have the property has child yeah 中文初译: 3个或3个以上孩子中3个或者3个或3个以上孩子。 。 。 00:47:14.250 EN: and where the set of these has child properties is larger or equal than three so that's um a number restriction so and in the same way we can not only build larger or equal we can also build a smaller or equal so we could here say for example parents with less than three children less or equal than three children or less or equal two for example um 中文初译: 并且如果这些孩子的共有子女财产大于或等于3个,那么这是一个数字限制,所以和以同样的方式,我们不仅可以建造更大的或平等的子女财产,我们也可以建造一个较小的或平等的子女财产,这样我们可以在这里说,例如,子女少于3个或少于3个或少于3个或少于2个或等于2个(例如)子女的父母。 00:47:46.150 EN: and then we basically select all individuals which have the property r in that case only less or equal n times and then equal is the same so we select those individuals which have the property r exactly n times so these are the unqualified number restrictions why unqualified 中文初译: 然后,我们基本上选择所有拥有该财产的人, 在这种情况下,只有较小或相等的n 倍,然后等同是相同的。 所以我们选择那些拥有该财产的个人, 准确的 n 倍,所以这些是绝对数量限制 为什么不合格 00:48:15.790 EN: because we don't um say um what the value of these properties are needs to be from yeah so we don't indicate that the children needs to be belong to i don't know small to the class of concept of small children or adult children or something like this yeah that's unqualified doesn't matter just the number of properties um or tuples um belonging to the interpretation counts that's why 中文初译: 因为我们不表示这些属性的价值是什么 需要从是 是,所以我们不表示 孩子们需要属于我 00:48:47.090 EN: unqualified there also exists qualified number restrictions but we don't discuss them here that's a more complex uh description logics and the corresponding our constructs are our mini cardinality our max cardinality and our cardinality you remember these were all restrictions um and you had to these were several triples not just one triple you had to define the cardinality you had to define on what property uh 中文初译: 也有一些合格的数量限制 但是我们在这里不讨论这些 更复杂的 呃描述逻辑 而相应的 我们的构思是 00:49:17.750 EN: so that were two or three triples you needed to define such cardinality restrictions in our good then nominals another feature um and this is called o um a l c o if we add also nominals what is nominals um if we have a set of individuals a one to a n a nominal a one to a n is a concept which semantics is defined 中文初译: 也就是说,你需要2到3三倍的 来定义这种基本限制 在我们的商品中,然后标出另一个特性 嗯,这被称为Oum a l co 00:49:50.290 EN: as follows so we basically take the um the interpretation um is just the set itself yeah so we define a concept by naming all the individuals belonging to the concept we define a class by naming all the instances of the class you remember we had this um i think moons of mars yeah and they were we were just listing the moons of mars and in that way defining 中文初译: 因此,我们基本上采取 um(嗯) 解释 um(嗯)只是设置本身 是的,所以我们定义一个概念 通过命名所有属于 我们定义一个类别的概念的人 00:50:19.650 EN: the concept moon of mars yeah the owl construct is one of all one of so that our nominals yeah so this is basically a lot of features which can be added to alc to them our simple description logic um so and if we add all these features together so role functionality transitivity role hierarchy inverse roles 中文初译: 火星的月亮的概念 火星的月亮。 猫头鹰的构造是所有其中之一 这样我们的标语就是 是的,所以这基本上是许多特征 00:50:48.810 EN: and so on we come to a complex relatively complex description logic we started with the very simple one right in the beginning and we added now a lot of features to our description logic we also always saw how the interpretation of the logic is constructed or can be traced back to the set theoretical constructs and then we can do reasoning on top and there are three types of reasoning uh deductive reasoning uh 中文初译: 因此,我们来到了一个复杂的相对复杂的描述逻辑 我们从一开始就以非常简单的一个 开始,现在我们增加了许多特征 在描述逻辑中,我们还经常看到 如何构建或追溯到对逻辑的诠释 00:51:21.350 EN: deductive reasoning starts with the assertion of a general rule and proceeds from there to a guaranteed specific conclusion from the general rule to the specific application that's deduction then we have the opposite is induction it begins with observations that are specific and limited in scope and proceeds to a generalized conclusion that is likely but not certain in light of accumulated 中文初译: 推论从主张一般规则开始,然后从一般规则开始,从一般规则到有保证的具体结论,从一般规则到具体应用的保证具体结论,从一般规则到具体应用的保证具体结论,而具体应用的推论则是扣减,我们则相反的是引论,它从具体和范围有限的意见开始,进而得出根据累积情况可能但不确定的普遍结论 00:51:51.090 EN: evidence so from the specific to the general that's inductive reasoning and finally we have up the abductive reasoning it begins with an incomplete set of observations and proceeds to the likeliest possible explanation for the set so now how can we do logical reasoning in description logics and here you have a small um illustration so penguins are black 中文初译: 所以从具体到一般的证据, 从具体的到一般的 暗示推理 最后我们有了 绑架推理 它从不完整的一组观察开始 00:52:22.630 EN: and white some old tv shows are black and white therefore some penguins are all tv shows yeah uh so logic is another thing that penguins aren't very good at uh so um yeah logic is um can become quite quite complex and um if you if your knowledge base um contains some um omissions 中文初译: 白色一些旧的电视节目是黑白的 所以有些企鹅都是电视节目 是的,嗯,所以逻辑是另一回事 企鹅不善于做 呃,嗯... ... 00:52:51.190 EN: or some mistakes yeah um you will quickly identify and that happens all the time if you build a more complex knowledge base you will definitely make mistakes and you will figure out that there are some errors and the knowledge representation and reasoners can actually help you identifying these errors and then you can fix them you can come to a more precise knowledge representation 中文初译: 或一些错误 是的,嗯 你会很快地发现 而且只要你建立一个更复杂的知识库 你肯定会犯错 你会发现有些错误 00:53:15.670 EN: and um yeah that is sometimes crucial for example these kind of things are applied a lot in life sciences so in life sciences there are really large ontologies and knowledge bases and what do you need them for you need them for for example for advisory systems for physicians in the hospitals yeah so that when a patient has certain symptoms that you could infer what 中文初译: 嗯,是的,有时这非常关键 比如说 生命科学中大量应用这类东西 生命科学中 确实有大量的本科和知识基础 00:53:44.550 EN: could be the disease the patient is suffer from suffering from yeah and uh this of course needs to be correct and that's why reasoning is applied a lot in life science application in other areas reasoning is not so important for example i showed you also schema.org and and google's knowledge graph i think google's knowledge graph is not very consistent there are some mistakes and errors it doesn't really matter 中文初译: 病人可能患的疾病 是,是的。 当然,这当然需要正确。 这就是为什么为什么在生命科学应用中 在其他领域应用大量推理 应用了很多生命科学应用 00:54:16.530 EN: if there is something not 100 correct there because and it's not life-threatening yeah and i guess many of you have experienced that already that also some information you get from google's knowledge graph is wrong or you use the navigation from a to b or the characteristics of some restaurant and maybe the opening hours were not correct or something else so there are knowledge graphs which do not need to be 100 correct 中文初译: 我想你们中很多人已经经历过 谷歌知识图上的一些信息也是错误的 或者使用从a到b的导航 或某些餐馆的特征 也许开放时间不正确 或者其他东西 所以有些知识图不需要100对 00:54:45.850 EN: but there are also knowledge graphs which need to be 100 correct and for example life sciences is such an area where it should be in many cases 100 correct and that's why reasoning for example is applied a lot in that area so logical reasoning and description logics is done let's if you have an interpretation i and t and t box a and a box 中文初译: 但也有一些知识图表 需要100正确。 例如生命科学是一个领域, 在很多情况下应该100正确。 这就是为什么举例说的理由在这个领域应用了很多。 所以逻辑推理和描述逻辑就完成了。 如果你有一个解释i、t和t框a和一个方框的话。 00:55:14.970 EN: and k a knowledge base which consists of t and a we say then i is a model for t for our terminology box if i if all the axioms alpha hold in i for every axiom alpha in the t box and then we write that the interpretation i fulfills the t box t so it's denoted in the way yeah 中文初译: k 一个知识库 由 t 和 a 我们说, 那么i 是一个我们的术语框的t 模式, 如果如果所有 exicoms alpha 都保留在i i 中, 对于t 框中的每个 exicoom alpha 来说, 然后我们写 解释i 满足 t box t , 所以用这个方式来表示是的 00:55:52.630 EN: so it needs to be fulfilled for all individual axioms and then it also holds for the whole set of axioms t similarly i can be a model for the a box um if it holds for every a box axiom alpha and the same way and then we can say that i is also a model for the whole knowledge base if it's a model for the t box and the a box yeah very simple 中文初译: 所以它必须满足所有个体的轴轴, 然后它也保持整个的 轴轴(t类似地) 我也可以做一个盒子的模型 如果每个箱子的正exom alpha 都以同样的方式 00:56:23.330 EN: and then we can say an axiom alpha is entailed by our knowledge base k written in the following way k entails alpha if every model i of k is a model for alpha yeah so we can infer new axioms alpha if every model every interpretation of k 中文初译: 然后我们可以说一个正正正角阿尔法 是由我们的知识库K产生的 k 以以下方式写 K 包含字母 如果每个k的模型i 都是一个阿尔法的模型 那么我们可以推断新的正正正正正方形 alpha 00:56:54.060 EN: is also a model for alpha that's the way how reasoners work then there are some reasoning services so the first one is concept satisfiability um so what does it mean we want to check whether a concept c is satisfiable with regard to our knowledge base k so whether there exists a model i of k such 中文初译: 也是一种阿尔法模式, 也就是理性者如何工作的方式 然后有一些推理服务 所以第一个是概念相对性 嗯 那么我们想检查一个c概念相对于我们的知识库是否可作比较 k, 那么是否存在这样的模型i, k 00:57:23.960 EN: that the concept is not empty yeah so we want to see um or we want we want that c equals the button concept if that does not hold in our knowledge base yeah then it's satisfiable when if c is not always the button concept it can actually be that you define a class but the class is not satisfiable um 中文初译: 概念不是空的。 是的, 所以我们想看到 um 或者我们想要我们想要的c 等同按钮概念 如果它不在我们的知识库中存在的话, 那么它就是可允许的 当如果 c 不总是按钮概念 它实际上可能是你定义一个类, 但是该类是不可坐的 uh 00:57:58.750 EN: can we look at some example so for example you can define uh for example no you can define lots of classes for example if you define the intersection of parents and not parents yeah that will be always um then this will be not satisfiable if you define parents as having the has child property for example and not parent 中文初译: 举例来说,我们可以看看一些例子 比如你可以定义,例如 例如无法定义很多类。 比如,如果你定义父母的交叉点而不是父母的交叉点。 是的,这总是 um 。 那么如果你把父母定义为拥有孩子财产(比如)而不是父母 00:58:27.940 EN: of having not they has child property when you then build the intersection um the intersection will be always empty right so it's easy to define concepts which cannot be fulfilled yeah you can come up yourself with some examples as an exercise um and um sometimes this happens by accident of course in this simple example which i showed you now it's very obvious yeah 中文初译: 当你建起十字路口的时候,他们没有孩子财产。 那么十字路口就永远是空的, 这样很容易定义那些无法完成的概念。 你可以自己举一些例子来做一个练习 嗯,嗯,有时这当然是偶然发生的。 这个简单的例子我现在告诉你了它非常明显。 00:58:56.120 EN: that if i have a class defined as parents and not parents but in some cases it's not so obvious to see that a class actually cannot contain any instances and the reasoner can identify this because this usually hints to a modeling mistake because of course we don't want to define empty classes we want to define concepts and classes which have instances so if a reasoner finds out that certain concepts are not satisfiable 中文初译: 如果我有一个被定义为父母而不是父母的类别,但在某些情况下,看到一个类别实际上不能包含任何实例并不那么明显。 理智者可以识别这一点,因为通常这暗示了建模错误,因为我们当然不想定义空类,我们想界定概念和类别,如果一个理性者发现某些概念不容许的话,我们想要界定这样的概念和类别。 00:59:25.580 EN: that helps us to figure out what mistakes we have done in the modeling right another reasoning service the second one here is subsumption and basically here a reasoner checks um the problem of checking whether c is subsumed by d with regard to our knowledge base so we can check whether the set of individuals belonging to the concept c is a subset of 中文初译: 这有助于我们找出我们在右建模中犯了什么错误 在另一个推理服务中 我们做了哪些错误 我们做另一个推理服务 在这里的第二个是自食其力 00:59:55.780 EN: the individuals belonging to the concept d in every model interpretation I of K. Yeah, so we can basically check if some classes are subclasses of each other. 中文初译: 在对K的每一个模型解释I. 是的,所以我们基本上可以检查某些类别是否是彼此的子类。 01:00:08.460 EN: Sometimes they might not be defined as subclasses, but we might figure out they are actually subclasses, yeah. So, for example, you can define the classes of males and females and humans and then you figure out, oh, the class of males and the females are actually subclasses of the class of humans. 中文初译: 有时它们可能不是亚类, 但我们可能会发现它们其实是亚类。比如说,你可以定义 男性、女性和人类的类别, 然后你就会发现, 哦, 男性和女性的类别实际上是人类类中的子类。 01:00:32.340 EN: And that a reasoner can tell you and discover such subclass relationships. So that's also very helpful. And it sometimes also helps you to identify also mistakes in your ontology, because if you discover, oh, an animal is a subclass of a human, then apparently something is wrong, yeah. 中文初译: 理性人可以告诉你并发现这种子类关系。所以这也非常有用。有时它也帮助你识别你的本体论中的错误, 因为如果你发现了,哦,动物是人类的子类, 那么显然有些东西是错误的,是的。 01:00:54.960 EN: That's something which is unintended and the last reasoning service here on this slide is satisfiability. So consistency, the problem of checking whether a knowledge base K is consistent, for example, or whether it has model, whether it is possible actually to create a model for this knowledge base. 中文初译: 这是意料之外的事情, 这张幻灯片上最后的推理服务是可比较性。 所以一致性, 比如说, 检查知识库K是否一致的问题, 或者它是否有模型, 能否为这个知识库创建一个模型。 01:01:28.500 EN: Yeah, that's satisfiability and of course we want to create models or we want to create knowledge bases which have models. 中文初译: 是的,这是可鄙性 当然我们想创造模型 或者我们想创造有模型的知识基础 01:01:36.580 EN: If it's not satisfiable, it also usually tells us there are some mistakes in our knowledge base. When the knowledge base is inconsistent, something is wrong. 中文初译: 如果它不是可讽刺的, 它通常也告诉我们我们的知识库中有一些错误。 当知识库不一致时, 就会有问题。 01:01:51.950 EN: Then we have instance checking another reasoning service. So we can basically check whether an instance belongs to a class, yeah, whether A belongs to the concept C here in that case, whether this holds, this axiom holds in our knowledge base K. 中文初译: 然后我们通过实例检查另一个推理服务。 所以我们基本上可以检查一个例子是否属于某个类别, 是的,A是否属于C概念, 在那个情况下,这个概念是否属于C, 这个轴线是否在我们的知识库K中存在。 01:02:09.910 EN: So for every model I of K, so for example, you might not, so this, for example, also happens when you have the subsumption reasoning, when you have defined somebody as a man or a woman, then a reasoner will determine if men and women are subclasses of humans, right, that an individual, 中文初译: 所以对于每一个K的模型I, 比如说,你可能不会, 因此举例来说,这也会发生。 当你有补贴推理的时候, 当你把某人定义为男人或女人时, 当你把一个人定义为男人或女人的时候, 那么一个理智者就会决定 男人和女人是否是人类的子类,对, 一个个体, 01:02:40.070 EN: Paul, for example, who is a man and Angela, who is a woman, that they are also humans, that they belong to the concept of humans. 中文初译: 例如,保罗是男人,安吉拉是女人,他们也是人,他们属于人类的概念。 01:02:49.850 EN: So this is instance checking what reasoners can do. Also retrieval, so we can use a reasoner a bit like a query engine to retrieve things. 中文初译: 这是实例检查理性者能做什么。 还可以检索, 这样我们就可以使用像查询引擎一样的理性者来检索东西 。 01:02:59.910 EN: So we can basically retrieve all instances A which belong to a concept C in our knowledge base. Yeah, so that's finding all individuals A which belong to a concept C with regard to K. 中文初译: 因此,我们基本上可以检索到知识库中属于C概念的所有 A 实例。是的,这就是找到所有属于K概念A的个人。 01:03:15.910 EN: So that's retrieval. It's a bit what, a bit similar to what a database engine does, right. The difference is for retrieval using a reasoner, the instance or the individual does not have to be explicitly assigned to a class. 中文初译: 这就是检索。它有点什么, 有点类似于数据库引擎的作用,对。区别在于 使用理智的检索, 案例或个人不必被明确指定为类。 01:03:32.510 EN: For example, if you want to retrieve all humans, it will, a reasoner will retrieve automatically all males and all females, if males and females were defined as subclasses of human, even if the individuals itself were not defined explicitly as belonging to humans. 中文初译: 例如,如果你想找回所有人类,就会有理性者自动找回所有男性和女性,如果男女被定义为人类的子类,即使个人本身没有被明确界定为属于人类。 01:03:55.570 EN: So it would here retrieve Angela, Paul and a lot of others, even if they were not explicitly defined to be humans, but just males and females. 中文初译: 因此,它会在这里找回安吉拉、保罗和许多其他人, 即使他们不是被明确定义为人类, 而是只指男性和女性。 01:04:07.110 EN: Yeah, and that's what a reasoner can do. This example is of course relatively simple, but you can imagine if you have complex class hierarchies and if you have restrictions, then retrieval is not so simple and can become very complicated. 中文初译: 对,理性人可以这样做。这个例子当然比较简单, 但是你可以想象如果你有复杂的等级结构, 如果你有限制,那么检索就不那么简单, 并且变得非常复杂。 01:04:23.710 EN: And another reasoning service is realization, and this is the problem of finding all named classes C which an individual A belongs to with regard to our knowledge base K. 中文初译: 另一种推理服务是实现, 这就是如何找到所有C类名字的问题。 A类个人属于K类知识库。 01:04:38.210 EN: So find all C for a given A such that if that A is an element of the interpretation of C in every model I of K. 中文初译: 所以找到给定A的全部C 如果A是K的每个模型I中 C解释的一个要素 01:04:53.450 EN: Yeah, so we want for an individual to know to which classes it belongs. For example, if we have Angela and maybe Angela is explicitly said to be a woman, but then the realization will tell us it's also a human because womans are a subclass of humans. It's a living being because humans are a subclass of living beings and it's an all thing. 中文初译: 是的,所以我们想让个人知道它属于哪一类。例如,如果我们有安吉拉, 也许安吉拉被明确地说是一个女人, 但是这个认识会告诉我们它也是人, 因为女人是人类的子阶级。它是活物,因为人类是生命的子阶级, 它是所有事物。 01:05:25.610 EN: Everything is an all thing, so all thing will always be returned here as a named class. So these were reasoning services and there are a few more. 中文初译: 一切都是一回事,所以所有的事情 都会以一个命名的阶级被送回这里。所以这些是推理服务,还有几个。 01:05:38.170 EN: We can reduce all services to satisfiability check. That's very important. So a reasoner actually only does one thing. 中文初译: 我们可以将所有服务都减少为可视性检查。 这非常重要。 所以理智者实际上只做一件事。 01:05:45.490 EN: He does a satisfiability check. And then you can reduce the other reasoning services to this satisfiability. So concept satisfiability. 中文初译: 他检查了是否合理。然后你可以将其他推理服务 降低到这种合理性。所以概念是否合适。 01:06:00.410 EN: So we want to check if the axiom C equal bottom concept does not hold. So we can just check does there exist an X such that K union C X is satisfiable. 中文初译: 所以我们要检查xiom C 等值底部概念是否有效。 所以我们可以检查是否存在 X 这样Kail CX 是值得嘲笑的。 01:06:21.290 EN: So we add or we look for some new individual X and then we add the axiom X belongs to the concept C to our knowledge K graph K or knowledge base K and we check whether this is satisfiable. 中文初译: 因此,我们增加或者寻找一些新的单个 X 然后我们把xiom X 属于C的概念加入到我们的知识K图K或知识库K中 然后我们检查这是否值得争议。 01:06:41.070 EN: If it's satisfiable, then the concept C is also satisfiable and this does not hold here. Subsumption. So how can we realize subsumption? 中文初译: 如果这是可比较的,那么C概念也是可比较的, 并且它不能维持在这里。 补贴。 那么我们如何实现可计算? 01:06:56.890 EN: So we want to check whether C is subsumed by D and what can we do? We just check whether we add to our knowledge base K the following axiom C and conjunction complement D of X and we check whether this is unsatisfiable. 中文初译: 所以我们要检查C是否被D吸收了,以及我们能做些什么?我们只是检查一下我们是否在知识库中添加了 K 的以下正则 C 和 连结 补充 D 的 X, 我们检查这是否不能令人满意。 01:07:24.150 EN: So if this is unsatisfiable, then the subsumption must hold. So if there is no X which belongs to the complement of D and to and to C then this cannot cannot be then it must be a subsumption. 中文初译: 因此,如果这是无法满足的,则必须维持这种补贴。因此,如果没有属于D和C补充物的X,那么,这不可能是不能满足的,那么它必须是一种补贴。 01:07:53.770 EN: So you can also visualize this as a small exercise as a fan diagram. So draw basically the sets of C D and the complement of D and then you can see that the subsumption only holds in the case when this does not hold. 中文初译: 所以您也可以将它想象成一个小练习,作为风扇图。所以基本上绘制 CD 和 D 的组合,然后可以看到子吸收只有在不维持的情况下才有效。 01:08:16.190 EN: And similarly instance check. Checking whether something is an instance. So we just add the complement to our knowledge graph and check knowledge base and check whether it's unsatisfiable. 中文初译: 类似实例检查。 检查某物是否是一个实例。 所以我们只需在知识图表中添加补充内容, 检查知识库, 检查是否无法满足。 01:08:31.370 EN: So a reasoner only has to provide one reasoning service and that's satisfiability and with this one you can do all the other six which we have actually shown here. 中文初译: 所以一个理性者只需要提供 一种推理服务, 那就是相对性。 通过这个 你可以完成我们在这里实际展示的 所有其他六种方法。 01:08:43.930 EN: Instance checking, retrieval, realization, subsumption, concept satisfiability. They can all be traced back to a satisfiability check. 中文初译: 程序检查、检索、实现、子假设、概念可控性。它们都可以追溯到对等性检查。 01:08:57.090 EN: Good and how does the reasoner now do that? For that there is the tableau algorithm and the tableau algorithm helps to prove the satisfiability of a concept. 中文初译: 以及理智者现在如何做到呢?因为有排列算法和排列算法有助于证明概念的相对性。 01:09:11.090 EN: So remember a concept is satisfiable if there exists a model i satisfying it. Now we need to construct a decision procedure for constructing models and that's exactly what the tableau algorithm does. 中文初译: 所以请记住,如果存在一种我满足它的模式,那么一个概念是可讽刺的。现在我们需要为构建模型构建一个决定程序,这正是排列算法的作用。 01:09:26.630 EN: And the procedure is we transform a given concept into a negational normal form also called NNF. Then we apply completion rules in as long as possible and the concept is satisfiable if and only if a clash-free tableau can be derived to which no completion rule is applicable. 中文初译: 我们的程序是把一个特定的概念变成一个又称为NNF的否定的正常形式。 然后我们尽可能长时间地适用完成规则,只有在能够产生一个没有完成规则的无冲突桌的情况下,这个概念才值得嘲讽。 01:09:50.170 EN: So this clash-freeness is important. We will see in a minute how that works. So let's look at some example. A sample ontology for tableau algorithm. So we have a T-box. 中文初译: 因此这种无冲突状态很重要。 我们将在一分钟内看到它是如何运作的。 让我们来看看一些例子。 一种表单算法的样本本体论。 所以我们有一个T框。 01:10:06.750 EN: Man is defined to be the complement of woman intersected with person. Woman is a subclass of person. Mother is defined as a woman which has the child property, at least one child property. 中文初译: 男性是指与人交织在一起的妇女的补充,女性是个人的一个子类,母亲是指拥有孩子财产、至少一个孩子财产的妇女。 01:10:26.150 EN: That's our T-box. Yeah man is a bit complicated defined here as not a woman intersect with person but that's how it is here. 中文初译: 是的,男人有点复杂 这里的定义不是女人和人交错 而是这里就是这样 01:10:37.930 EN: And then we have the A-box. We have Stephen is a man, Monica is not a man, Jessica is a woman and we have Stephen and Jessica and Stephen has a child Jessica. 中文初译: 然后我们有了A盒。我们有了Stephen是个男人,Monica不是男人,Jessica是女人 我们有了Stephen和Jessica 还有Stephen有一个孩子,Stephen有Jessica 01:10:52.330 EN: Maybe instead of woman we should have used female here male and female because Jessica apparently is a child depending on how old she is. 中文初译: 也许我们不应该用女人,而应该用女人 因为Jessica显然是个孩子 取决于她多大年纪 01:11:01.950 EN: She might not yet be a woman but doesn't matter. This terminological finesse is here. So let i be an interpretation with. 中文初译: 她可能还不是女人,但没关系。这个术语学的精细图在这里。所以让我来解释一下。 01:11:13.030 EN: So we have man, the concept man and we have Stephen in there. Woman we have then Jessica and Monica in there. Mother we have Monica in as an interpretation of the concept mother. 中文初译: 所以,我们有男人,概念人和史蒂芬在那里。 女人,然后杰西卡和莫妮卡在那里。母亲,我们有莫妮卡作为概念母亲的解释。 01:11:31.370 EN: The interpretation of person then corresponds to Jessica, Monica and Stephen and has child Monica, Stephen, Stephen, Jessica. 中文初译: 然后,对人的解释相当于Jessica、Monica和Stephen,并有孩子Monica、Stephen、Stephen、Jessica。 01:11:46.070 EN: So apparently Monica is the grandmother of Jessica here in that case right. So Monica has a child Stephen and Stephen has a child Jessica. 中文初译: 很明显Monica是Jessica的祖母,所以Monica有一个孩子Stephen, Stephen有一个孩子Jessica。 01:11:58.290 EN: So then holds this interpretation here fulfills our T-box. Yeah and it also fulfills our A-box. Yeah so we have an interpretation. 中文初译: 然后在这里保持这种解释 满足了我们的Tbox。是的,它也满足了Abox。所以我们有一个解释。 01:12:13.130 EN: Interpretation basically means defining the sets the concept or assigning sets to the concept interpretations. Yeah and here this interpretation holds. 中文初译: 解释基本上意味着界定各套概念,或指定各套概念解释。 01:12:28.690 EN: You can look carefully you will find that everything like fits for example has child. We have Stephen and Jessica here right. 中文初译: 你可以仔细看一看,你会发现所有像样的东西都有孩子。 01:12:38.350 EN: It's in there but there are also different interpretations possible. This particular interpretation here for example defines Monica to be the mother of Stephen but there could be also a different interpretation. 中文初译: 它在那里,但也有不同的解释。比如说,这个解释把Monica定义为Stephen的母亲, 但也可能有一个不同的解释。 01:12:56.770 EN: And now we need this negational normal form. And the negational normal form, a concept is a negational normal form if all occurrences of negatives are in front of atomic concepts. 中文初译: 现在我们需要这种否定的正常形式。而否定的正常形式, 一种概念是一种否定的正常形式, 如果所有负面现象都出现在原子概念面前的话。 01:13:12.150 EN: Yeah so every ALC concept can be transformed into an equivalent one in negational normal form using the following rules. 中文初译: 因此,每一个ALC概念都可以按照下列规则,以否定的正常形式转变为等同的概念。 01:13:21.330 EN: Yeah so if we only have C then the negational normal form of C is C if C is atomic right. So if we have the complement of C then the negational normal form is complement C if C is atomic right. 中文初译: 所以,如果我们只有C的话,那么如果C是对原子的话,否定的正常的C形式就是C。所以如果我们有C的补充,那么如果C是对原子的话,否定的正常形式就是C的补充。 01:13:41.070 EN: So it's then still directly in front of an atomic concept here. Yeah so then we have the negational normal form of the double complement of C is the same as the negational normal form of C. 中文初译: 因此,它仍然直接站在原子概念的面前。是的,所以我们有了C的双倍补充的否定的正常形式 和C的否定的正常形式一样。 01:13:56.990 EN: The negational normal form of C conjun, disjunction D is the negational normal form of C. Disjunction negational normal form of D. The same with the conjunction here yeah. 中文初译: 否定的正常形式的C conthern,分离的D是否定的正常形式的C. D.脱离的正常形式。 01:14:18.770 EN: Then we have the negational normal form of the complement of a disjunction. And that is then the negational normal form of so we basically take the negation inside yeah. 中文初译: 然后,我们有了一种否定的正常形式 来补充脱节。然后,这就是否定的正常形式, 所以我们基本上把否定的内容放在里面。是的。 01:14:33.790 EN: So that's the negational normal form of the complement of C. And then we turn the disjunction into a conjunction yeah. 中文初译: 这就是C的正常补充的否定形式 然后我们把分离变成一个组合 是的 01:14:47.710 EN: So we you see this is also here turned around yeah because the negation from outside can be pulled inside if we turn the conjunction into a disjunction. 中文初译: 所以,我们可以看到这里也在这里 转过身来。 是的,因为从外部的否定可以被拉到里面 如果我们把结合变成分裂。 01:15:02.010 EN: And vice versa the conjunction is turned into a disjunction here yeah. We can also pull the negation inside here to build the negational normal form. 中文初译: 反之亦然,连结也变成了一种脱节。我们也可以在这里拉动否定的内涵 来构建否定的正常形式。 01:15:17.340 EN: And then we have the negational normal form for all what is for all role restrictions. So there we basically take the negational normal form inside right. 中文初译: 然后我们拥有所有角色限制的否定性正常形式。 因此,我们基本上在右侧 采取否定性的正常形式。 01:15:31.180 EN: We apply it on the concept C. The same for the extension. We also apply it to the concept C because the negational normal form just tries to push negations inside to the atomic concepts yeah. 中文初译: 我们把它应用到C概念上, 扩展也一样。我们也把它应用到C概念上, 因为否定的正常形式只是试图将否定推入原子概念中去。 01:15:47.260 EN: That's why here nothing needs to be changed because we don't have a negation in there. If we have a negation around we actually have to switch the all value restriction into an existential value restriction. 中文初译: 这就是为什么这里没有什么需要改变的, 因为我们没有否定。 如果我们周围有否定的话, 我们实际上必须把所有的价值限制 转换成存在的价值限制。 01:16:01.960 EN: And then we can push the negation down to the concept inside yeah. And vice versa with the negation of an extension restriction here we can push that inside by converting it to an all value restriction yeah. 中文初译: 然后我们可以把否定推到概念内部去。是的,反之亦然。反之,我们也可以把它转换成所有价值限制 来将扩展限制推到内部去。 01:16:22.600 EN: So these are the rules and you can apply them if you have a complex concept expression you can apply these roles as much as you want and then you basically in the end get the negation in normal form. 中文初译: 这就是规则,你可以应用这些规则, 如果你有一个复杂的概念表达方式, 你可以尽可能地运用这些角色, 然后基本上最终得到正常的否定。 01:16:44.320 EN: So for example we have this concept here. We have a negation of a conjunction B, this negation B, this junction negation C. Yeah and then what what do we do? 中文初译: 比如说,我们这里就有这个概念。 我们否定了B连线、B连线、B连线、C交汇点,是的,然后我们该怎么办? 01:17:02.260 EN: We build a negational normal form and then we apply the the rules yeah step by step. So first we convert this the negation of this conjunction into a disjunction of the negations right. 中文初译: 我们形成了一种否定的正常形式, 然后我们一步一步地应用规则。 所以首先我们将这种否定的结合 转换成对否定的权利的分离。 01:17:20.980 EN: That's why we add here a negation here the disjunction here a double negation. So then we can resolve the double negation then it becomes only C here and here we can resolve the double negation then it only becomes A disjunction negation B. 中文初译: 这就是为什么我们在这里添加一个否定, 此处的矛盾是双重否定。 这样我们就可以解决双重否定, 然后它就变成C了。 在这里我们可以解决双重否定, 然后它就变成了B了。 01:17:39.780 EN: So the negational normal form of C is C because it's an atomic concept. The negation normal form of A conjunction complement of B is then the negation a normal form of A conjunction negation normal form of complement of B disjunction C sorry disjunctions yeah and then the negation a normal form of A as an atomic concept is just 中文初译: 因此,C的正反常形式是C, 因为它是一个原子概念。那么,A的正反常形式是B的正反常形式,而A的正反常形式则是A的正反常形式, 与B断开C的正常互不相容补充形式。 是的,然后否定A作为原子概念的正常形式就是 01:18:09.520 EN: A negation a normal form of complement of B is let's look back here in our rules yeah negation a normal form of complement of C is just a complement of C if C is atomic and since in our case here B is atomic yeah so and then we have the negation normal form is this one A disjunction complement B disjunction C that's the negational normal form 中文初译: 否定B的正常补充形式 就是让我们从我们的规则中回顾一下。 是的,否定C的正常补充形式只是C的补充,如果C是原子的话。 因为就我们的情况而言,B是原子 是的。 因此,我们有否定的通常形式是这个