answer
Thing is special, in description logic it corresponds to TOP, so every class should imply Thing. Literally you can say Thing means everything.
An instance of Thing is kind of the universal instance, however I never created an instance of Thing before.
The initial domain of a property is certainly Thing, so there is no need to remove Thing from the domain unless you restrict the domain to a specific subclass of ThingThanks a lot!
Another question: If I am going to use ontology to facilitate biology data mining, would it be better to create two ontologies (biology, and data mining) rather than one (universal framework of bioDataMining)?
To what extent Instantiating the ontology to instances will benefit my framework? Or can I implement it without much instances (to write my thesis)
I know this is a abstract question, still hope I can get some hints.
Building ontology is really tough since you can build one in so many ways. If your domain is big enough or the domain can be deemed as several modules each of which has a distinct concentration , it should be better to split the domain into several small ontologies. It cause no harm since you can import any part as long as you wish to. However, it is really up to the ontology engineer, in this case it's you, to decide which is the better way. You'd better take a look at some ontological engineering books to get a feel about this project.
As for the instantiation, it is still up to you
以下是引用fishinfire在2007-3-30 16:59:00的发言:我们知道owl的属性有三种:object datatype annotation。其中annotation属性是注释,我们可以不去管他。另外两种属性中,object属性的语义表达能力很强,可以通过对object属性的限制制定关于类的等价公理。但是不知道datatype属性能不能进行限制呢? 举个例子说,我想定义食肉动物,只要定义两条限制就可以制定了食肉动物这个类的等价公理,即:1、是animal的子类;2、eat allvaluefrom animal。这是object属性的典型应用。 当我想定义成人(区别于儿童)这个类时,定义了一个datatype属性age(整形),想描述如下限制:1、是human的子类;2、年龄大于等于18。第二条怎么描述呢? 我想OWL应该能表达这种很简单的逻辑关系,但我不知道它是如何做到的。
敬请指教。
关于datatype可以看官方说明:http://www.w3.org/TR/owl-guide/#Datatypes1
现在的OWL对于你说的“年龄大于等于18”没有很好的表示方式,因为OWL里头只可以说明DATATYPE的值是多少,也就是说有“等于”,却没有“大于”或者“小于”。。。。。。
要想推理这个,你就要自己把数值1-18都做nominal,然后判断某个实例的hasAge的range是不是在这个NOMINAL类里头。。。
我们上次就做了1-100个nominal用来推理人的年龄,还好大多数人都活不过100岁。。。晕。。。
knowledge representation-->declarative knowledge representation-->non-logic (network-based) declarative knowledge representation--> description logic...
以下是引用tainechill在2007-3-29 11:36:00的发言:最近看一些本体/语义方面的论文时,看到一些 is_a, part_of, whole_of 的关系表达符号,请问这些符号来自于哪个语言或框架模式。。
如下例,两个类分别由 owl:class 来描述,而之间的领域内的关系 presents 由 owl:objectProperty 来描述。 这些都应该是 owl lite 里的 constructor
Service See comments above - - ServiceProfile See comments above
...
There are no cardinality restrictions on this property. http://www.daml.org/services/owl-s/1.2/Service.owl#Service" /> http://www.daml.org/services/owl-s/1.2/Service.owl#ServiceProfile" /> http://www.daml.org/services/owl-s/1.2/Service.owl#presentedBy" />
以下是引用jl1022在2007-4-10 20:48:00的发言:父类子类的关系不能用于表示partof 关系,原因很明显, 人的一部分“手”能继承“人”的“有智慧”这个属性吗?显然不能。其实我看到很多论文里面都犯了这个原则性错误,用subclass 来表示 partof, 我认为这种低级错误不应该出现在发表的文章里面,有的甚至是博士论文里面。比如在某博士论文里面把“磨头”这个“磨床”的部件作为“磨床”的一个子类。 我就纳了闷了,这么低级的错误难道他们都看不出来吗?还将他发表到自己的博士论文里面。还是我的认识有错误,想听听大家的看法
An instance of Thing is kind of the universal instance, however I never created an instance of Thing before.
The initial domain of a property is certainly Thing, so there is no need to remove Thing from the domain unless you restrict the domain to a specific subclass of ThingThanks a lot!
Another question: If I am going to use ontology to facilitate biology data mining, would it be better to create two ontologies (biology, and data mining) rather than one (universal framework of bioDataMining)?
To what extent Instantiating the ontology to instances will benefit my framework? Or can I implement it without much instances (to write my thesis)
I know this is a abstract question, still hope I can get some hints.
Building ontology is really tough since you can build one in so many ways. If your domain is big enough or the domain can be deemed as several modules each of which has a distinct concentration , it should be better to split the domain into several small ontologies. It cause no harm since you can import any part as long as you wish to. However, it is really up to the ontology engineer, in this case it's you, to decide which is the better way. You'd better take a look at some ontological engineering books to get a feel about this project.
As for the instantiation, it is still up to you
以下是引用fishinfire在2007-3-30 16:59:00的发言:我们知道owl的属性有三种:object datatype annotation。其中annotation属性是注释,我们可以不去管他。另外两种属性中,object属性的语义表达能力很强,可以通过对object属性的限制制定关于类的等价公理。但是不知道datatype属性能不能进行限制呢? 举个例子说,我想定义食肉动物,只要定义两条限制就可以制定了食肉动物这个类的等价公理,即:1、是animal的子类;2、eat allvaluefrom animal。这是object属性的典型应用。 当我想定义成人(区别于儿童)这个类时,定义了一个datatype属性age(整形),想描述如下限制:1、是human的子类;2、年龄大于等于18。第二条怎么描述呢? 我想OWL应该能表达这种很简单的逻辑关系,但我不知道它是如何做到的。
敬请指教。
关于datatype可以看官方说明:http://www.w3.org/TR/owl-guide/#Datatypes1
现在的OWL对于你说的“年龄大于等于18”没有很好的表示方式,因为OWL里头只可以说明DATATYPE的值是多少,也就是说有“等于”,却没有“大于”或者“小于”。。。。。。
要想推理这个,你就要自己把数值1-18都做nominal,然后判断某个实例的hasAge的range是不是在这个NOMINAL类里头。。。
我们上次就做了1-100个nominal用来推理人的年龄,还好大多数人都活不过100岁。。。晕。。。
knowledge representation-->declarative knowledge representation-->non-logic (network-based) declarative knowledge representation--> description logic...
以下是引用tainechill在2007-3-29 11:36:00的发言:最近看一些本体/语义方面的论文时,看到一些 is_a, part_of, whole_of 的关系表达符号,请问这些符号来自于哪个语言或框架模式。。
如下例,两个类分别由 owl:class 来描述,而之间的领域内的关系 presents 由 owl:objectProperty 来描述。 这些都应该是 owl lite 里的 constructor
...
以下是引用jl1022在2007-4-10 20:48:00的发言:父类子类的关系不能用于表示partof 关系,原因很明显, 人的一部分“手”能继承“人”的“有智慧”这个属性吗?显然不能。其实我看到很多论文里面都犯了这个原则性错误,用subclass 来表示 partof, 我认为这种低级错误不应该出现在发表的文章里面,有的甚至是博士论文里面。比如在某博士论文里面把“磨头”这个“磨床”的部件作为“磨床”的一个子类。 我就纳了闷了,这么低级的错误难道他们都看不出来吗?还将他发表到自己的博士论文里面。还是我的认识有错误,想听听大家的看法
0 Comments:
Post a Comment
<< Home