宋炜博士是英国杜伦大学(University of Durham)副教授，主要从事Semantic Web
相关研究和教学工作(http://www.dur.ac.uk/w.w.song/)。宋老师自2002年开始在
北京大学计算机系开设“元数据，本体和语义网——语义网高级课程”，并与北京大
学张铭老师合著了第一本中文Semantic Web专著《语义网简明教程》（2004年6月出版）。

宋老师做客W3CHINA.ORG讨论区Semantic Web版（http://bbs.w3china.org/sw.asp
）的时间是 7月23日（周五）下午2:30-4:30。

为了大家能够和宋老师交流的方便和惬意，请大家自觉遵守如下几点小小的要求：

* 请大家发贴尽量在本贴中，回贴时尽量使用引用，以免由于多个话题，引起混乱。

* 请大家在与宋老师交流的过程中中英文均可，但尽量用英文表达一些术语。

* 请大家在与宋老师讨论或向宋老师请教问题时尽量围绕宋老师的研究方向（即
Semantic Web）展开话题，以保证在有限的时间内达到最好的讨论效果。

* 考虑到参加讨论的人可能比较多，提的问题也会很多。尽管时间不允许宋老师一一
作出回答，但宋老师会在其中挑选出一些最具代表性的问题给予回答。

这是一次非常好的与高水平学者交流的机会，欢迎有兴趣的朋友在约定的时间登
陆W3CHINA.ORG讨论区加入到与讨论中来。

首先，非常感谢宋老师百忙之中能够抽出时间来这里做客。我在2002年听过您的讲课，受到很多启发。首先我想描述一下我们目前工作的基本思路：

我们计划开发的是中国电子政务ontology，然后利用这些ontology来描述电子政务资源。带有了ontology描述信息的电子政务资源管理平台可以做很多现在很难做到的事情。

我们在开发电子政务ontology的过程中，发现了这样的一些问题,主要还是构建ontology的一些问题。提出来，想请教宋老师。

A. 问题边界难以确定；
    
如果仅仅是针对某个应用，根据需求来开发ontology的话，还好办一些。但问题是开发电子政务这样的领域ontology，涵盖的主题非常广，想把它做成一个电子政务领域通用的知识本体。在某个具体应用需求不是非常清楚的情况下，要求Ontology能够细化到一个什么样的程度，很难把握。宋老师能否给出一些建议。

B.领域交叉之后的本体融合问题

电子政务本身也不是一个完全独立的领域，电子政务信息中可能还会涉及到电子商务、农业、教育等等一系列的其它领域知识。如果这些领域也都相应地开发了电子商务本体、农业本体、教育本体；那么就不可避免会面对一个本体融合的问题。这些本体之间重叠的部分如何处理好？感觉比较困难。

C.Ontology的管理和运营

Ontology作为今后智能信息化建设的基本砖块，可能会有越来越多的团体和组织加入到ontology的开发中来。为了避免ontology开发过程中的混乱和无序的状态，我认为需要有一个ontology的开发、注册、查找、维护、检查、管理中心。这个中心包括了整套的运营机制和技术实现。但是目前看到的ontology 注册中心实现的文章比较少，宋老师能否推荐一些资料。

谢谢宋老师！

1。[求助]关于OWL语义推理的例子 http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8889

2。Jumbo：  OWL系统  http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8940
各位高手，请问现在有没有比较成型的或是已经在试用的用owl写的系统或是模拟系统

呢？如果有，望告知网址或参考资料，谢了

3。 作者：xinba2002 http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8829
  哪位对基于语义WEB的知识管理系统有研究，请指点一二！能推荐一些研究实例最好

4。 [求助]一个小问题：为什么是rdf:Property，而不是rdfs:Property[/URL]
http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8754

谢谢！

I am very pleased to be given this chance to discuss SW issues. Although I have been working in the area for a number of years, as it grows very fast, I think my knowledge is very limited and however I wouold like to share with you.

Regarding your questions, they are related to a large and unstable domain - e-gov.
Simply speaking, your understanding of the domain knowledge comes from the domain experts. What you need to do is to use ontology languages to formulate the domain knowledge. It does not matter that matter whether the problem domain is complete or not. You can always let them open.

diversity of ontology is a good issue. Maybe that ius why reasoning is introduced. I will continue this discussion after I send this out first. So I am known on.

以下是引用trevol在2004-7-23 13:58:53的发言： 虽然应该2：30开始，但是心情激动，早早就来了。。。先提些问题。。。呵呵 首先，非常感谢宋老师百忙之中能够抽出时间来这里做客。我在2002年听过您的讲课，受到很多启发。首先我想描述一下我们目前工作的基本思路： 我们计划开发的是中国电子政务ontology，然后利用这些ontology来描述电子政务资源。带有了ontology描述信息的电子政务资源管理平台可以做很多现在很难做到的事情。 我们在开发电子政务ontology的过程中，发现了这样的一些问题,主要还是构建ontology的一些问题。提出来，想请教宋老师。 A. 问题边界难以确定；       如果仅仅是针对某个应用，根据需求来开发ontology的话，还好办一些。但问题是开发电子政务这样的领域ontology，涵盖的主题非常广，想把它做成一个电子政务领域通用的知识本体。在某个具体应用需求不是非常清楚的情况下，要求Ontology能够细化到一个什么样的程度，很难把握。宋老师能否给出一些建议。 B.领域交叉之后的本体融合问题 电子政务本身也不是一个完全独立的领域，电子政务信息中可能还会涉及到电子商务、农业、教育等等一系列的其它领域知识。如果这些领域也都相应地开发了电子商务本体、农业本体、教育本体；那么就不可避免会面对一个本体融合的问题。这些本体之间重叠的部分如何处理好？感觉比较困难。 C.Ontology的管理和运营 Ontology作为今后智能信息化建设的基本砖块，可能会有越来越多的团体和组织加入到ontology的开发中来。为了避免ontology开发过程中的混乱和无序的状态，我认为需要有一个ontology的开发、注册、查找、维护、检查、管理中心。这个中心包括了整套的运营机制和技术实现。但是目前看到的ontology 注册中心实现的文章比较少，宋老师能否推荐一些资料。 谢谢宋老师！

I am still maintaining the issue that for a application/p[roblem domain we have a number of ontologies, each deal with a particular aspect of the problem. The inter-ontology relationships are maintained by e.g. semantic network or other means. In this way, we will make the ootology pure "holding with only one relation among the concepts described in the ontology."

For you last question, I am not sure whether there are good/recommended ontolgoy  especially for e-gov. I would suggest you to take a look at semanticweb.org and gils standard.

以下是引用w.w.song在2004-7-23 14:40:33的发言： First of all, sorry about using English as I find it not easy to type Chinese using my keyboard. I would take this opportunity to say thanks to all of you for your initerest in the very promising research area. I am very pleased to be given this chance to discuss SW issues. Although I have been working in the area for a number of years, as it grows very fast, I think my knowledge is very limited and however I wouold like to share with you. Regarding your questions, they are related to a large and unstable domain - e-gov. Simply speaking, your understanding of the domain knowledge comes from the domain experts. What you need to do is to use ontology languages to formulate the domain knowledge. It does not matter that matter whether the problem domain is complete or not. You can always let them open. diversity of ontology is a good issue. Maybe that ius why reasoning is introduced. I will continue this discussion after I send this out first. So I am known on. [quote]以下是引用trevol在2004-7-23 13:58:53的发言： 虽然应该2：30开始，但是心情激动，早早就来了。。。先提些问题。。。呵呵   首先，非常感谢宋老师百忙之中能够抽出时间来这里做客。我在2002年听过您的讲课，受到很多启发。首先我想描述一下我们目前工作的基本思路：   我们计划开发的是中国电子政务ontology，然后利用这些ontology来描述电子政务资源。带有了ontology描述信息的电子政务资源管理平台可以做很多现在很难做到的事情。   我们在开发电子政务ontology的过程中，发现了这样的一些问题,主要还是构建ontology的一些问题。提出来，想请教宋老师。   A. 问题边界难以确定；          如果仅仅是针对某个应用，根据需求来开发ontology的话，还好办一些。但问题是开发电子政务这样的领域ontology，涵盖的主题非常广，想把它做成一个电子政务领域通用的知识本体。在某个具体应用需求不是非常清楚的情况下，要求Ontology能够细化到一个什么样的程度，很难把握。宋老师能否给出一些建议。   B.领域交叉之后的本体融合问题      电子政务本身也不是一个完全独立的领域，电子政务信息中可能还会涉及到电子商务、农业、教育等等一系列的其它领域知识。如果这些领域也都相应地开发了电子商务本体、农业本体、教育本体；那么就不可避免会面对一个本体融合的问题。这些本体之间重叠的部分如何处理好？感觉比较困难。   C.Ontology的管理和运营   Ontology作为今后智能信息化建设的基本砖块，可能会有越来越多的团体和组织加入到ontology的开发中来。为了避免ontology开发过程中的混乱和无序的状态，我认为需要有一个ontology的开发、注册、查找、维护、检查、管理中心。这个中心包括了整套的运营机制和技术实现。但是目前看到的ontology 注册中心实现的文章比较少，宋老师能否推荐一些资料。      谢谢宋老师！ [/quote]

Only an answr to your small question: rdf:Property instead of rdfs:Property came from a requirement where most of the modelling primitives shoudl be dealt with in RDF and leave more space for RDFS to deal with terminology (ontology) issues.

以下是引用orangebench在2004-7-23 14:26:42的发言： 版上的一部分没有得到很好解决的问题，可以在本贴中或原始贴中讨论： 1。[求助]关于OWL语义推理的例子 http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8889 2。Jumbo：  OWL系统  http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8940 各位高手，请问现在有没有比较成型的或是已经在试用的用owl写的系统或是模拟系统 呢？如果有，望告知网址或参考资料，谢了 3。 作者：xinba2002 http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8829    哪位对基于语义WEB的知识管理系统有研究，请指点一二！能推荐一些研究实例最好 4。 [求助]一个小问题：为什么是rdf:Property，而不是rdfs:Property[/URL] http://bbs.xml.org.cn/dispbbs.asp?boardID=2&ID=8754 谢谢！

1。[opmissing]
假如要构建的本体主要是用来推理的，那么对于这个这样本体有什么特殊的要求吗？

2。  [npubird]
我有两个小问题，一直没有个清晰的理解。希望能得到各位同学的帮助。
问题1：各种文献中总说到RDF(S)是一个Lightweight的本体语言，这怎么理解呢？
问题2：如何理解本体或者一个系统的封闭性呢？
谢谢！

我也来描述一下我目前的工作及思路，请宋老师指点。

我的硕士论文题目是关于Service Matching的。主要是想寻求一种非逻辑方法（比如soft computing）作为逻辑方法的补充，以支持soft matching。关于soft matching，我看过一些paper，但和我的想法不太一样。

我觉得semantic web service和database中的soft matching，最大的不同就是SWS有ontology-based service description，这使得我们可以在进行"soft" matching的同时尽最大可能保证匹配结果在语义上的有效性（简单的说，就是保证匹配质量）；而数据库的的soft matching，完全是通过computing进行的，这样就难免会得出一些absurd的结果。

请问宋老师，有没有可能将 soft computing 技术结合 ontology，用于服务的soft matching？

谢谢！

I have read some of your questions published. I am happy to see that many are quite interested in the ontology building and KM. As one step toward the Semantic Web, ontolgoy has displayed its powerful support in describing concepts, in particular, in certain applications domains. However, in other areas, such as e-gov, health care, e-learning, etc. to build up a good ontology is still very difficult as the problems are so diversal and heterogeneous.

There are a number of ontology languages. But the languages do not tell us how to acquire domain informaiton and knowledge. So what we can do is to start from a very small sub-sub-problem, use a specific realtionship (IS-A), manually design an ontology (maybe in natural language at beginning). Then test it in the reality. When you have a number of these, I believe you will find out something you really need.

The Government Information Locator Service (GILS) is an effort to identify, locate, and describe publicly available Federal information resources, including electronic information resources. GILS records identify public information resources within the Federal Government, describe the information available in these resources, and assist in obtaining the information. GILS is a decentralized collection of agency-based information locators using network technology and international standards to direct users to relevant information resources within the Federal Government.

Because this collection is decentralized, the GPO Access GILS site is only one of several US Federal GILS sites. Thirty-two agencies have mounted their GILS records on the GPO Access server (and are listed individually in the scroll box located on the [URL=http://www.access.gpo.gov/su_docs/gils/index.html]GILS search page[/URL]). The GILS records of the other departments and agencies are spread across many sites. For basic information about GILS, visit [URL=http://frwebgate.access.gpo.gov/cgi-bin/leaving.cgi?from=exithosted.html&to=http://www.gils.net&log=linklog]www.gils.net[/URL]. One of the Federal GILS sites that cannot be accessed via GPO Access is that of the Department of Defense, whose GILS records can be searched at [URL=http://frwebgate.access.gpo.gov/cgi-bin/leaving.cgi?from=exithosted.html&to=http://sites.defenselink.mil&log=linklog]http://sites.defenselink.mil.[/URL]

For agencies that mounted their GILS records on a server other than GPO Access, GPO wrote GILS pointer records, to describe their GILS holdings and link to those holdings. GPO sent those pointer records to each agency and invited each to edit its own records as appropriate. [URL=http://www.access.gpo.gov/su_docs/gils/agency-pointer.html]Browse Pointer GILS[/URL] records.

The Government Printing Office partially funded a study entitled, [URL=http://www.access.gpo.gov/su_docs/gils/gils-eval/index.html]"An Evaluation of the Federal Government's Implemantation of the Government Information Locator Service: The Final Report"[/URL], which is available on GPO Access. This study was completed on June 30, 1997.

以下是引用w.w.song在2004-7-23 14:46:22的发言： I continue this discussion. I am still maintaining the issue that for a application/p[roblem domain we have a number of ontologies, each deal with a particular aspect of the problem. The inter-ontology relationships are maintained by e.g. semantic network or other means. In this way, we will make the ootology pure "holding with only one relation among the concepts described in the ontology." For you last question, I am not sure whether there are good/recommended ontolgoy  especially for e-gov. I would suggest you to take a look at semanticweb.org and gils standard.                               ~~~   [/quote]

赞成宋老师的观点。。。。

2. RDF is not an ontology language. RDFS is. But it maintains only subclassof and subpropertyof, which are basic ones for an ontology.

3. For this question, I am not sure whether I really see your meaning. Current ont languages follow RDF(S) and maintain extensibility. So the languages' contructors do not put any restrictions on objects. Should not be considered closure. However, when reasoning is introduced, to cetain extent, the corresponding sets should be closed for operations. Otherwise, it will cause the problem of uncertainty, etc.

以下是引用orangebench在2004-7-23 14:55:31的发言： 呵呵，再转两个问题： 1。[opmissing] 假如要构建的本体主要是用来推理的，那么对于这个这样本体有什么特殊的要求吗？ 2。  [npubird] 我有两个小问题，一直没有个清晰的理解。希望能得到各位同学的帮助。 问题1：各种文献中总说到RDF(S)是一个Lightweight的本体语言，这怎么理解呢？ 问题2：如何理解本体或者一个系统的封闭性呢？ 谢谢！

2. RDFS is lightweighted if you consider whether it contains logic operations like in your example. That is right. However, RDFS was designed to pave a road toward ontology. It is open-ended. DAML, OWL are built on top of it.

以下是引用npubird在2004-7-23 14:55:52的发言： 宋老师您好。 我有两个小问题请教： 1。在Semantic Web环境下，如何理解一个系统的封闭性呢？封闭的系统能比开放的系统带来哪些好处呢？ 2。对于RDF(S)，各种文献中说到它是一种Lightweight（国内翻译为轻量级）本体语言。对于这个说法，理解为：RDF（S）比较简单，不足以对某些知识进行表示，如不能表示“某本书只能有一个价格，但却至少有一个作者”这样的描述。这样的理解正确吗？

感谢宋老师和orangebech ！  gils 是个非常有价值的参考！

Maybe key issue lies in your semantics capture and representation instead of matching itself. How can you enbedded semantics in e.g. UDDI, or WSDL, etc.
right? Does this help?

以下是引用admin在2004-7-23 14:56:57的发言： 看到宋老师在线了~~~激动。。 我也来描述一下我目前的工作及思路，请宋老师指点。 我的硕士论文题目是关于Service Matching的。主要是想寻求一种非逻辑方法（比如soft computing）作为逻辑方法的补充，以支持soft matching。关于soft matching，我看过一些paper，但和我的想法不太一样。 我觉得semantic web service和database中的soft matching，最大的不同就是SWS有ontology-based service description，这使得我们可以在进行"soft" matching的同时尽最大可能保证匹配结果在语义上的有效性（简单的说，就是保证匹配质量）；而数据库的的soft matching，完全是通过computing进行的，这样就难免会得出一些absurd的结果。 请问宋老师，有没有可能将 soft computing 技术结合 ontology，用于服务的soft matching？ 谢谢！

以下是引用w.w.song在2004-7-23 15:08:56的发言： Currently, reasoning is based on subsumption. That means, an ontology should in most cases maintain IS-A relationship. I.e. the tree nodes are connected in terms of generalization. You may extend this to other relationships later, such as part_of. [quote]以下是引用orangebench在2004-7-23 14:55:31的发言： 呵呵，再转两个问题：   1。[opmissing]   假如要构建的本体主要是用来推理的，那么对于这个这样本体有什么特殊的要求吗？ [/quote]

宋老师，您对语义Web的研究和发展有什么看法？ 欧洲在语义Web的研究力量是比较强大的，您在英国感觉到别人对SW的看法是怎样的？

我已进入博士理论研究阶段，并对网格已有一段时间的研究。现在，对Semantic Web很感兴趣，想了解Semantic Grid领域目前的研究进展和趋势，谢谢！

以下是引用npubird在2004-7-23 15:04:32的发言： 再问一个问题： 宋老师，现在SW的研究中对Ontology有这样一致的看法：建构一个通过的Ontology不现实，Semantic Web上将存在很多小的Ontology，某些是针对某个行业的，某些是针对某个应用的，某些是权威的机构发布的，某些是个人发布的，某些是知识专家开发的，某些是普通用户开发的。 这样一来，如何处理这些众多的本体的关系呢，如何管理和利用它们呢？如果Web上的Ontology越来越多，越来越重要，是不是会改变人们对当前的Web的架构的看法呢？

我们大部分都是在读的研究生，对语义Web感兴趣，想做这方面的研究，但苦于能给指导的人并不多（纯粹个人猜测，如果不符，抱歉），宋老师，能否给些关于SW研究的建议。 谢谢！

；－）

我问的是有关XML智能查询（Intelligent Search on XML Data）和SW的问题，
是不是XML智能查询可以用SW技术实现，比如将XML Schema映射成本体
这个过程是不是很复杂？
映射成本体之后怎么办？
（我刚刚接触这方面的东西，能否多给些指导）
thanks！

能否再请宋老师谈谈对WSMO的印象？它是否会比OWL-S有希望？

另外能否请您比较一下Semantic Grid Service与SWS这两个比较相近的概念。他们在service description and matching 方面有什么异同。 谢谢！

以下是引用w.w.song在2004-7-23 15:24:17的发言： Was writing but suddenly gone. Well, I have a student doing this kind of work. You may go to Carnegie Mellon University and Toshiba and check their work on SWS matching. Maybe key issue lies in your semantics capture and representation instead of matching itself. How can you enbedded semantics in e.g. UDDI, or WSDL, etc. right? Does this help?

以下是引用orangebench在2004-7-23 15:24:29的发言： 如果本体要用来推理的话，除了要有基本的IS-A关系，等基本的表达能力外，它的推理复杂度应该不要太高，另外，所用的本体语言应该是可判定的，所以，我觉得OWL Lite是个不错的折中。 [quote]以下是引用w.w.song在2004-7-23 15:08:56的发言： Currently, reasoning is based on subsumption. That means, an ontology should in most cases maintain IS-A relationship. I.e. the tree nodes are connected in terms of generalization. You may extend this to other relationships later, such as part_of.   [quote]以下是引用orangebench在2004-7-23 14:55:31的发言：   呵呵，再转两个问题：    1。[opmissing]    假如要构建的本体主要是用来推理的，那么对于这个这样本体有什么特殊的要求吗？   [/quote] [/quote]

What I can see in the future, SW will go close to AI, Semantic Grid/e-Sci. go close to KM.

以下是引用orangebench在2004-7-23 15:31:34的发言： 我问一个大一点的问题： 宋老师，您对语义Web的研究和发展有什么看法？ 欧洲在语义Web的研究力量是比较强大的，您在英国感觉到别人对SW的看法是怎样的？

以下是引用orangebench在2004-7-23 15:40:36的发言： 再来一个类似的大大的问题： 我们大部分都是在读的研究生，对语义Web感兴趣，想做这方面的研究，但苦于能给指导的人并不多（纯粹个人猜测，如果不符，抱歉），宋老师，能否给些关于SW研究的建议。 谢谢！ ；－）

我的问题是：
1. 如何将跨越众多领域的分类法和主题词表用现在的本体方式来组织和表示？这样的目的是将原来由图书馆员来完成的资源分类组织和管理，能够通过计算机来方便地处理和完成，效率更高。

2. 未来数字图书馆中的检索是否可以完全使用知识查询(搜索)的方式来完成？在实施方法与途径上，能否请宋老师指点指点？

我们的本体是西方哲学的本体论思想在人工智能、知识表示等领域的体现。
但在哲学上，很多学者的在反思本体论后，
逐渐认为ontology的研究不可能得到结果。
因此逐渐转向认识论、价值论的研究发展。
例如维特根斯坦的“家族类似”观点，认为本质属性并不存在。
回到我们这边的本体来，此时，我们建立的各种各样的本体，
将因为个体例外的存在，
使本体推理难以保证结果的正确性。

您认为，我们的本体是否会存在上述问题？

谢谢！

以下是引用w.w.song在2004-7-23 15:57:19的发言： I think the breakthrough of SW will be either (or both) in Semantic Grid (e-Sci.) and Web services, where Web services will be closely realted to Grid. One of the SG research activities is virtual organization (VO). Another is SGS (Semantic Grid Service). Fundamental research issues include 1) how to capture semantics for Grid nodes and 2) how to integrate these nodes to form a service or organization. One of the current problem is Web service lanagueges are at low level while SW language at high. Key should be to generate a THING in between. [quote]以下是引用grid在2004-7-23 15:35:30的发言： 宋老师，您好！   我已进入博士理论研究阶段，并对网格已有一段时间的研究。现在，对Semantic Web很感兴趣，想了解Semantic Grid领域目前的研究进展和趋势，谢谢！ [/quote]

以下是引用lxtahu在2004-7-23 15:40:37的发言： 宋老师您好！ 我问的是有关XML智能查询（Intelligent Search on XML Data）和SW的问题， 是不是XML智能查询可以用SW技术实现，比如将XML Schema映射成本体 这个过程是不是很复杂？ 映射成本体之后怎么办？ （我刚刚接触这方面的东西，能否多给些指导） thanks！

我也非常关注 MICKY 的问题

以下是引用w.w.song在2004-7-23 16:17:26的发言： In general, XML documents do not really provide any semantics, so intelligence has nowhere to stand there. As you mentioned, a mapping from XML or XMLS to RDF or ont may help. As I know, there some researchers at PKU try this issue already. Real semantic mapping is not easy but you can always try a light weighted method, like partially semantic mapping. I had thought about this recently, to share with you, I would suggest, if applicable, to make a parallel document with your XMLS, called annotation document, where conceptual relationships between XML elements can be supplied. I am not sure whether this may give your some ideas. [quote]以下是引用lxtahu在2004-7-23 15:40:37的发言： 宋老师您好！   我问的是有关XML智能查询（Intelligent Search on XML Data）和SW的问题，   是不是XML智能查询可以用SW技术实现，比如将XML Schema映射成本体   这个过程是不是很复杂？   映射成本体之后怎么办？   （我刚刚接触这方面的东西，能否多给些指导）   thanks！ [/quote]

以下是引用npubird在2004-7-23 15:59:40的发言： Semantic Grid我上个学期也看过一些资料，但能检索到的有用的文献不多，总的感觉现在提出的东西还比较粗，当时相关的网www.semanticgrid.org在很长一段时间没有更新内容。最终我还是放弃了对这个方向的进一步探索。 目前国内已经有学校申请到了Semantic Grid的相关研究基金。 IEEE Intelligent System也开辟了相关的Semantic Grid的专期。 我也特别想听听宋老师对Semantic Grid的看法。 [quote]以下是引用grid在2004-7-23 15:35:30的发言： 宋老师，您好！   我已进入博士理论研究阶段，并对网格已有一段时间的研究。现在，对Semantic Web很感兴趣，想了解Semantic Grid领域目前的研究进展和趋势，谢谢！ [/quote]

WE HAVE TO STOP HERE.
THANK YOU VERY MUCH FOR THE DISCUSSIONS AND VERY GOOD QUESTIONS.

以下是引用micky在2004-7-23 16:08:48的发言： 宋老师，在您的《语义网简明教程》中提到“数字图书馆作为下一代因特网资源管理的核心技术，...”。这两年数字图书馆的概念已经提升到基于知识概念体系的资源管理和知识管理，在各种领域各种规模范围内都得到长足的发展和应用。数字资源的语义元数据加上分类法和主题词表(本体)等就构成为一个庞大的知识库。 我的问题是： 1. 如何将跨越众多领域的分类法和主题词表用现在的本体方式来组织和表示？这样的目的是将原来由图书馆员来完成的资源分类组织和管理，能够通过计算机来方便地处理和完成，效率更高。 2. 未来数字图书馆中的检索是否可以完全使用知识查询(搜索)的方式来完成？在实施方法与途径上，能否请宋老师指点指点？

//最崇高的敬意！

以下是引用w.w.song在2004-7-23 16:35:10的发言： WE HAVE TO STOP HERE. THANK YOU VERY MUCH FOR THE DISCUSSIONS AND VERY GOOD QUESTIONS.

让我们用热烈的掌声感谢并欢送宋老师！

以下是引用w.w.song在2004-7-23 16:35:10的发言： DL (digital library) is indeed playing an important role introducing metadata to web resources. Lib catalog plus current search method (data) is a good KB. I think this method may be good for organizing web information in some way. However, lib catalog is not purely an ont. Many other domain ont are quite different from lib catalog. Integration is a problem. Lib catalog is neither a common sense ont, like what what we have in our mind. I don't think DL methods will dominate the Web or SW. Sorry my guest has arrived. WE HAVE TO STOP HERE. THANK YOU VERY MUCH FOR THE DISCUSSIONS AND VERY GOOD QUESTIONS. [quote]以下是引用micky在2004-7-23 16:08:48的发言： 宋老师，在您的《语义网简明教程》中提到“数字图书馆作为下一代因特网资源管理的核心技术，...”。这两年数字图书馆的概念已经提升到基于知识概念体系的资源管理和知识管理，在各种领域各种规模范围内都得到长足的发展和应用。数字资源的语义元数据加上分类法和主题词表(本体)等就构成为一个庞大的知识库。   我的问题是：   1. 如何将跨越众多领域的分类法和主题词表用现在的本体方式来组织和表示？这样的目的是将原来由图书馆员来完成的资源分类组织和管理，能够通过计算机来方便地处理和完成，效率更高。   2. 未来数字图书馆中的检索是否可以完全使用知识查询(搜索)的方式来完成？在实施方法与途径上，能否请宋老师指点指点？    [/quote]

以下是引用hawk在2004-7-23 16:10:07的发言： 宋老师，您好。 我想请教这样一个问题。 我们的本体是西方哲学的本体论思想在人工智能、知识表示等领域的体现。 但在哲学上，很多学者的在反思本体论后， 逐渐认为ontology的研究不可能得到结果。 因此逐渐转向认识论、价值论的研究发展。 例如维特根斯坦的“家族类似”观点，认为本质属性并不存在。 回到我们这边的本体来，此时，我们建立的各种各样的本体， 将因为个体例外的存在， 使本体推理难以保证结果的正确性。 您认为，我们的本体是否会存在上述问题？ 谢谢！

以下是引用grid在2004-7-23 16:16:23的发言： 谢谢！ 在Grid中，一切资源均以服务的形式向外提供，与Web services类似（现在Grid services与Web Services已出现融合的趋势）。请问宋老师，目前SW language是否足以描述特定领域的服务本体？如果从事该领域研究，难度会不会过大？ [quote]以下是引用w.w.song在2004-7-23 15:57:19的发言： I think the breakthrough of SW will be either (or both) in Semantic Grid (e-Sci.) and Web services, where Web services will be closely realted to Grid.   One of the SG research activities is virtual organization (VO). Another is SGS (Semantic Grid Service).   Fundamental research issues include 1) how to capture semantics for Grid nodes and 2) how to integrate these nodes to form a service or organization.   One of the current problem is Web service lanagueges are at low level while SW language at high. Key should be to generate a THING in between.   [quote]以下是引用grid在2004-7-23 15:35:30的发言：   宋老师，您好！    我已进入博士理论研究阶段，并对网格已有一段时间的研究。现在，对Semantic Web很感兴趣，想了解Semantic Grid领域目前的研究进展和趋势，谢谢！   [/quote] [/quote]

I would like to hear some comments from you on prospective key research issues in the semantic web (SW) field, and what you are going to choose to focus on? ;)

Here is some of my comprehension of the SW. From an application developer point of view, SW is not a big deal. It helps to solve some problem on describing things but mostly for interoperability sake. As we all face a step to describe data in any of our program, to be able to make it *universally* understandable, we tag it (which is the XML fashion). A more general case is to make a tag with another "tag" - namespace where one can refer to as a description of the tag. And if the namespace links to a URL document, it may be further described with ontology basis. So it ends up with that we are sharing model (probably partial) with others.

I think the key of SW is to share knowledge together with its model to others however with the premise that both share a common ontology underneath the model description. Given all these ready (in terms of documents and links), one can reason to get some results, such as a pear is a fruit which is a kind of thing to be rotted in 7 days at normal temperature during the autumn season on the east part of earth (remember this odd sentence from which old sci-film?:)).

But, this is really not a big deal as people have done huge work on knowledge representation so-called AI bubble failure before the recent IT and agent technology bubbles. One can easily borrow from them or modify to conform the SW formation e.g., from the earlier KIF form. Using existing reasoning engine (such as Jena using backward chaining reasoning). So in general behind the SW hype,  it is a set of tools (RDF, RDFS, OWL) to help to describe your data for a better interoperability whatever the underlying basis is (based on ontology or not) they can be treated as a blackbox. If these technologies survive, they will become common as today's XML and C language which everyone (programmer) is familiar with.

However, the problem is there is no one universal ontology to be able to describe common sense knowledge, so the use of ontology is limited to domain knowledge description, so everyone is going to define their ontologies for their models such as many by DAML and these applications will still be independent as they are nowadays until someday some commercial ontology (I would say a protocol) gets accepted by the market, then we will rely on it (isn't MS uPnP doing this connecting communication with semantic descriptions).

As SW technologies become *mature*, it is attractive for application developers who have new topic/standard to spend energy to conform to (just the matter of time for them) but really a dark day for researchers to repeat the failure road :(

My prediction of the extensions of near future SW (if it survives) willl be the revival of the old AI technology such as rule-based system with support for fuzzy logic, probablistic and uncertain reasoning, then machine learning things etc.

Thanks and just come to have a discussion here

/B.C.

Tim berners Lee 在他的design note里论述了他的理由：
1.A Semantic Web is not Artificial Intelligence
2.A semantic Web will not require every application to use expressions of arbitrary complexity
3. A semantic Web will not require proof generation to be useful: proof validation will be enough.
4. Knowledge Representation goes Global:
  We remove the centralized concepts of absolute truth, total knowledge, and total provability, and see what we can do with limited knowledge

但是，一直觉得他的认证并不充分，也一直没花时间去好好想这个问题。只能
说：相信W3C，相信Sir Tim Berners-Lee 吧。希望听听大家的想法。

参考：
Web design issues; What a semantic can represent
http://www.w3.org/DesignIssues/RDFnot.html

以下是引用beichuang在2004-7-26 20:45:10的发言： Dear Prof. Song, I would like to hear some comments from you on prospective key research issues in the semantic web (SW) field, and what you are going to choose to focus on? ;) ......... /B.C.

以下是引用beichuang在2004-7-26 20:45:10的发言： Dear Prof. Song, I would like to hear some comments from you on prospective key research issues in the semantic web (SW) field, and what you are going to choose to focus on? ;) Here is some of my comprehension of the SW. From an application developer point of view, SW is not a big deal. It helps to solve some problem on describing things but mostly for interoperability sake. As we all face a step to describe data in any of our program, to be able to make it *universally* understandable, we tag it (which is the XML fashion). A more general case is to make a tag with another "tag" - namespace where one can refer to as a description of the tag. And if the namespace links to a URL document, it may be further described with ontology basis. So it ends up with that we are sharing model (probably partial) with others. ..... Thanks and just come to have a discussion here /B.C.

So I think it is a good time for application developers, just make SW a plugable tech which can be used to extend the possibility of their applications, but not the key. So SW can be treated as a reasoning blackbox which can be replaced by any other technologies superior

/B.C.

以下是引用beichuang在2004-7-26 23:14:40的发言： I think it is a practical view of technology - just to see what will happen given technologies they defined, so there may be killer applications which will be accepted by users, however, we have to see whether it is because of the use of SW technology, like IM and VoIP etc, they do not use SW at all, however, SW may give some extra advantage while may not either. So I think it is a good time for application developers, just make SW a plugable tech which can be used to extend the possibility of their applications, but not the key. So SW can be treated as a reasoning blackbox which can be replaced by any other technologies superior /B.C.