- From: John Hardin <johnchardin@yahoo.com>
- Date: Mon, 15 Sep 2003 16:44:41 -0700 (PDT)
- To: Leo Sauermann <leo@gnowsis.com>, udef.builders@topica.com
- Message-ID: <2079984-1463792638-1063669483@boing.topica.com>
The following message was posted to the udef.builders mailing list: Leo, I replied and copied to the udef.builders group. Thanks for the clarifications. We are absolutely tracking toward the Semantic Web activity, due to the fact that there will always be differing names for identifcal data element concepts, and if the SW is to be automated, from a provisioning and integration standpoint, I think that the UDEF has a good answer. Actually, where you say that the URI based namespacing for the identifier doesn't offer any help for UDEF, on the contrary! I have long held the belief that eventually there will be a service ubiquitously available that will serve as a "semantic equalizer" in real-time. The service and consumer must either carry the semantic equivalency id within each's document tags, or refer to the namespace, as illustrated by you... thoughts anyone? Leo Sauermann <leo@gnowsis.com> wrote: thank you for your answer, I have had a look now on the mailing list, but I am so focused on RDF that I cannot participate, in our company resources are scarce at the moment as we are in developing a product on Semantic web. Management is fixed on that Looking at the mentioned prototypes and "proof of concepts" discuessed in the list, I assume that things are working good for UDEF. Perhaps UDEF and semantic web will integrate each other. It is easy to integrate UDEF in semweb: http://www.udef.org/idspace/1.1.1.1.1 could be used in semantic web, today, without any hassle. > When you refer to using DNS, are you talking about the protocol or the server / propogation network? The ultimate goal for this is a widely distributed (as you note, like DNS) reg/rep scenario, but with some specific semantic identification services (via SOAP and a user interface for domain experts to use in adding to the trees), change management (again, I agree with you on managing change) and other features. I think that outpaces the DNS applications. Well, I meant the practise we are using in semantic web / RDF applications: - The unique identities are urls - a url has two important parts: 1. the server hostname (unique through dns structure and "you have to buy it to use it") 2. the path on the host, uniqueness here has to be made by the developers and management behind. So what I meant with DNS is that the uniqueness of the server name is enough for constructing global uniqueness and this is reused. Certain bogus domains exist that are used sole for identifiers, f.e. http://ont.gnowsis.com/person might be used for identifiying a person in an ontology. and there is no need to run a host there. But this will not help UDEF in any way, as RDF uses strings as identifiers and UDEF structured numbers. Because you have the same needs for identifying things and work at a similiar scale, you will have to make a similiar structure like DNS has to run UDEF identification. A global network of primary servers, mirrors, protocol, management and social issues, companies selling and managing the space etc. That will perhaps make UDEF expensive to implement on a global scale. How we do the semantic mapping today is through 3 main ways: 1- embed rdf data as xml in normal http pages 2- have a link to rdf data as normal http service (like most RSS feeds are) 3- proprietary SOAP protocol or HTTP extensions like URIQUA did http://sw.nokia.com/URIQA.html but i know that you know these facts. I assume that in future, the two standards may reside side by side and copy from each other. For the Semantic Web community it is cakewalk to integrate the udef id's, so we will be to merge. greetings, Leo Sauermann Hi, I am developing RDF applications, beeing an experienced programmer, I create some kind of EAI on a personal level (www.gnowsis.com). I looked at the UDEF website and watched the http://www.udef.org/specdoc/UDEFv1pt02-Jan-2003_files/frame.htm I am a advocate for the semantic web and want to task some critical questions about UDEF and make some critical points from my point of view, I see some serious architectural flaws here. For my work I have done some research on Dewey and other categorization systems and think that the UDEF approach will surely fail. Technically: ======== You said that Dewey is "like DNS". Why don't you use DNS ? (the answer to this can only be politically motivated, all technical reasons speak for DNS. If you don't agree, you don't know DNS. Do you want to cash-cow companies with this ?) A successfull implementation of a dewey-similiar-system is the Network Management Protocol used to configure Routers and Switches: SNMP. But it shows that you will fail: - its proprietary. - it works only in the context of a handful of companies, because everybody has to have a numberspace assigned. Politically ======== try to get the major players on a table and decide on a numbering system. then manage CHANGES. Philosophically =========== RDF and the semantic web is built un the concept of LOCATORS, URL's. Tim Berners Lee was always opposed to the term URI, which just defines an "Identifier". Because: In a few years everything will be on the net and have a URL. We, the semantic web community, will push this movement. A vision is to bring everything on the net, addressable through an URL and secured through a web of trust. So every product you talk about, every order, every line in that order, every person, every ship, airplane, UPS parcel, car, pet, family, idea,. .... will be addressable through a URL. most of these things ARE ALREADY on the WEB ! You can see it start today in online shops or amazon. If you identify a book on your website, be honest, do you use ISBN or an URL to amazon ? This works so well because: - The institution that creates information has the means (technical) and the responsibility to publish their information on the web. So the creator of information has to provide the identifier. This is like in society, where you as creator name your project UDEF or your son "Frank" or your car "Herbie". If there is already a URL identifying concepts and physical objects, why have a parallel Identification system ? NET versus TREE ============= Tree structures far do not have the possibilities of web structures. In RDF you classify your information by linking it to other information. that is, the classification evolves through a web of linking. f.e.: The term "Person" that is needed in business was first defined in an ontology by the institution wordnet, in about 2000. Then Libby Miller and Dan Brickley from the FOAF group had also a concept of person, and subclassed it from the wordnet identifier (2001?). I myself do a work where, surely, are persons in my data space. (2003) So I subclass of FOAF or Wordnet. I use Wordnet, because I create a wrapper for MS-Outlook. -> What happens is that the most prominent identifiers are used very often and spread. The wordnet term was there first and it was good, was advertised, and so it spread and becomes a standard because we developers use it. Everything is created distributed and then interweaved through links like owl:sameIndividualAs and similiar. To illustrate this, you have to think about Darwin's theory of evolution: many different terms are created. then our "pop" and Internet culture happens: the top 3 movies (or cars, or company, or everything) are 10 times more popular than the rest in the top 10. All these effects have been widely discussed in the scientific community and are accepted. THIS WORKS. Now look what UDEF says: ====================== 1. Identify the applicable UDEF property word that characterizes the dominant attribute (property) of the data element concept. For example, Name, Identifier, Date, etc. 2. Identify the dominant UDEF object word that the dominant property (selected in step 1) is describing. For example, Person_Name, Product_Identifier, Document_Date, etc. 3. By reviewing the UDEF tree for the selected property identified in step 1, identify applicable qualifiers that are necessary to unambiguously describe the property word term. For example, Last Name 4. By reviewing the UDEF tree for the selected object identified in step 2, identify applicable qualifiers that are necessary to unambiguously describe the object word term. For example, Customer Person 5. Concatenate the object term and the property term to create a UDEF naming convention compliant name where it is recognized that the name may seem artificially long. For example, Customer Person Last Name 6. List common synonyms or business terms used within the context of a given industry. 7. Derive an intelligent UID based on the UDEF taxonomy that carries the UDEF inherited indexing scheme. For example <CustomerPersonLastName UID=“as.5_5.10”> ============================== This is by far not that elegant as the semantic Web. This is semantic Tree. Tree != Web. It is childish to think that all humans involved in the Knowledge Management process of a company will to these 7 steps and not fail. Step 1 involves decision that are not deterministic, everybody interprets the root terms differently, have you ever talked with some social scientist or philosopher about this ? I have, they ripped me apart and it was shaming how nifty they do it. So, whats the point in UDEF ? Why don't you hop on the semantic web wagon :-) regards, Leo Sauermann www.gnowsis.com as attachement is a fragment of a paper I write, a vision of myself about the future of the semantic web, to illustrate the way I am heading. -----Original Message----- From: www-rdf-interest-request@w3.org [mailto:www-rdf-interest-request@w3.org] On Behalf Of john hardin Sent: Saturday, September 13, 2003 11:23 PM To: smith@xmlhack.com; pfps@research.bell-labs.com; dave.beckett@bristol.ac.uk; a.allden@bristol.ac.uk; Libby.Miller@bristol.ac.uk; stefan@db.stanford.edu; studer@aifb.uni-karlsruhe.de; erdman@aifb.uni-karlsruhe.de; hendler@cs.umd.edu; ATO_Security@darpa.mil; guarino@loa-cnr.it; achille.varzi@columbia.edu; jernst@r-objects.com; mits1@umbc.edu; semanticweb@egroups.com; heflin@cse.lehigh.edu; seanl@cs.umd.edu; cgasarch@cs.umd.edu; lspector@hampshire.edu; acl@opus.cs.columbia.edu; bull_i3@univ-tln.fr; community@mlnet.org; daml-all@daml.org; dbworld@cs.wisc.edu; dl@dl.kr.org; info-ic@biomath.jussieu.fr; isworld@lyris.isworld.org; kaw@swi.psy.uva.nl; news-announce-conferences@uunet.uu.net; ontoweb-language-sig@cs.man.ac.uk; seweb-list@www1-c703.uibk.ac.at; ontoweb-list@www1-c703.uibk.ac.at; www-rdf-logic@w3.org; www-rdf-interest@w3.org; www-rdf-rules@w3.org; www-webont-wg@w3.org; dai-list@mcc.com; agents@cs.umbc.edu; topicmapmail@infoloom.com; iswc2003-list@www1-c703.uibk.ac.at; bernard.vatant@mondeca.com Subject: Open Letter to Web Services, Semantics, Semantic Web and Ontology Communities Dear Friends, Associates and Industry Leaders: This is an open letter, directed towards participants and leaders in the information technology communities who are focused on semantic-based data element identification, contextual representation (ontologies and taxonomies including frameworks for representing same) and communication of data element semantics within Web Services groups, the Semantic Web groups and e-business standards groups. We will not be frequently utilizing the list that was compiled for this introductory communication, but instead will compile a list of interested parties. Please indicate by reply if you are/aren't interested in receiving additional communications regarding this subject. Also, please forward this open letter to persons, organizations or publications that you believe will be interested. If this is a repeat, please excuse us, we will be eliminating duplicates in our mailings. ** Introduction ** The Aerospace Industry Association's Electronic Enterprise groups (http://www.aia-aerospace.org), the Association for Enterprise Integration (http://www.afei.org), the Electronics Industry Data Exchange (http://www.eidx.org) and others have been participating in the formation of an approach for identifying and managing semantic equivalency for data elements, and feel that this approach represents a potential approach for the IT industry, standards bodies and web services frameworks to move towards interoperability and convergence. This approach is termed the Universal Data Element Framework (http://www.udef.org), and is designed to provide a (semi) intelligent identifier, attached in some way to a data element (perhaps as an attribute within schemas or in an RDF based reference file), that can be resolved to produce an exact identification of the data element meaning. There is now a udef.builders mail list. Subscribe at udef.builders-subscribe@topica.com. View the archive at: http://www.topica.com/lists/udef.builders/read This open letter describes, at a high level, a summarization of the usability of the UDEF within the various semantics/ontology/taxonomy communities, and the current status and efforts around the UDEF. I am an active participant in the UDEF effort, and have been operating as co-chair of the AIA's UDEF committee. The opinions represented here are not necessarily those of the AIA http://www.aia-aerospace.org or AFEI http://www.afei.org who are the current stewards of the UDEF concepts and work efforts. I have recently left my last position as Practice Director over Enterprise Integrations at Envision, and am independent (http://www.geocities.com/johnchardin). As a result, for official UDEF information please contact Ron Schuldt, Lockheed Martin. You may reach him at ron.l.schuldt@lmco.com. I am continuing to work as a participant on UDEF efforts while seeking the next position, please feel free to contac t me with questions or comments at anytime. ** Groups represented in the Mail list ** There are a several groups with numerous activities devoted to the above stated issue. Some of the more prominent groups and activities which have been included on this mail are: - AIA's Electronic Enterprise Groups and Aerospace industry contacts - Logistics and Rail industry contacts - EIDX Technical Groups and Electronics industry contacts - Healthcare industry contacts, from pharmaceuticals and disease management co's - Open Application Group - Principals and the Semantics Working Group - RosettaNet Technical Dictionary and other contacts - ebXML Core Concepts groups - OASIS groups and OASIS UBL TC - WS-I groups - OWL and RDF groups - Semantic Web groups - TopicMap Groups - RosettaNet Dictionaries - Ontolog Forum - DAML-S - W3C xml:id (structural identifier for ontologies) - Basic Semantic Registry - Dublin Core - Software vendors: Contivo, Oracle, others - Various Industry standards bodies participants ** Usability Proposal ** There are a tremendous number of efforts on the net to establish the core stack for Web Services, including the basic protocols, security, multi-enterprise business process mechanisms and other items necessary for the vision of the Semantic Web and Automated Web Services to be a reality. Many of us are working on semantic approaches, including some unifying vocabularies such as UBL. However, there appears to be a need for basic, open, cross-industry specific identification of data element concepts so that all our efforts can communicate between one another to resolve the exact meaning (and perhaps context) of data elements. This is the target function for the UDEF, to provide a number/alpha based identifier that is easily represented in existing formats (XSD,DTD or other formats) as an attribute, or in structural mechanisms such as RDF. Few organizations trade and communicate only with organizations within thier industry. This presents a problem when partners are attempting to initialize e-business using two different formats: Which format is used? Which organization must remap the backend processes or middleware mappings to account for the new format? And, the question that the UDEF directly addresses: How do I know that your <POID> is the same data element concept as my <PurchaseOrderIdentifier>? Looking forward to the Semantic Web, when an intelligent agent representing me attempts to communicate my medical history to a doctor in relation to an appointment: How does the doctor's medical record system know that the data in <currentmedications> is the same as their systems' element labeled <patientpharmacology>? There are a multitude of industry standard XML ebusiness document formats, and some are emerging as horizontal across industries (OAGIS, RosettaNet and UBL are examples). Many are entrenched in specific industries (CIDX is a good example) and the application servers, messaging frameworks and backend mappings are already in use. In other words, the future of e-business may look like all companies utilizing a common business document format and dictionaries, but the immediate reality is that there are hundreds of formats in various depths of usage across the internet. We believe that during the course of e-business, there does (and will continue to for quite some time) exist a core requirement to map the semantically equivalent data element concepts between standards and other formats (RPC, document-literal or otherwise), e-business document formats, ontologies, taxonomies, RDF structures and data dictionaries. A universal resolvable semanitic identifier can provide some relief at design-time (read: mapping projects between partners or internally on enterprise integrations projects) by providing a semantic equivalency reporting mechanism. A pilot of this is described below in the Status section of this email. This can also, given the right framework of services, possibly assist in bringing about the real-time vision of the automated discovery, integration and execution of trading relationships, without placing a human analyst in the middle. Internationalization will also be greatly facilitated, by allowing a numbering format, instead of a language based format, as the pivot point in transformations on-the-wire. The metaphor that we have found works best to illustrate the function of the UDEF is that of the Dewey Decimal System: The UDEF attempts to label data elements in thier formats, as books are labeled in libraries, in a common, repeatable, resolvable, open way. The Dewey numbering system provides an intelligent identification and organizing capability within every library that uses it. (Caveat: this metaphor doesn't carry well outside the US, where schemes other than the Dewey Decimal System identify published works, but you get the point....) I believe that we are at critical mass with this issue in the internet technical community, and would like to propose a collaboration of the efforts of the various groups included in this communication, to form perhaps an over-arching discussion and/or working group(s) that can further the work for the usage scenarios, architectural frameworks (including propogation of core trees across the net, much as the DNS network propogates), data element matrix publishing and sharing (including public vs. closed vs. private availability of UDEF labeled document references for standards formats and software vendors data dictionaries, etc.), and other necessary tasks to bring up a dynamic network of semantic identification nodes. Perhaps this convergence will work into a W3C or OASIS (or both) sponsored recommendation spec. Please take the time to review the UDEF materials that are on the UDEF web site, and comment via the forms on the site, or to Ron Schuldt (ron.l.schuldt@lmco.com), or myself (johnchardin@yahoo.com) with ideas, comments, criticism, and reality checks. Or call me directly at the number listed below. Domain Experts: Please consider involvement in the upcoming effort to build out data elements contained in the trees with data elements reflecting domain expertise, slated to begin Q1 2004. If you wish to be placed on the maillist, or participate as a domain expert, in pilots, communicate ideas to the group, or anything else, please email myself johnchardin@yahoo.com and Ron Schuldt ron.l.schuldt@lmco.com . I extend this introductions and these suggestions with a great amount of respect to the participants in the various standards bodies. We are all working to stand up innovative, disruptive technologies and frameworks that are bringing about a new economy and great opportunity for us all. Best regards, and thank you for so much of your time today. John C Hardin 314.749.0562 mobile http://www.geocities.com/johnchardin (advance apologies for the popups from geocities) ** Current Status and Progress on UDEF Issues ** The perseverance of Ron Schuldt (15 years since the conception of the UDEF approach) is paying off, along with the efforts of a wide variety of groups and persons (thanks to you all). There is currently strong support or high levels of interest among the following groups: Industry Associations and eBusiness Groups - Aerospace, Defense, Electronics, Computer industries all providing support - Automotive, Airline, Railroad interested Private and Public Companies - Lockheed Martin, Boeing, Raytheon, Tyco Electronics, Envision, Ambrosoft and AMS are supporting - Northrup Grumman, DoD Logistics and Korean Ministry of Defense are interested Software Vendors - Contivo is supporting - Oracle, IBM, Microsoft, SeeBeyond, DataDirect, GXS and Unicorn are all interested Standards Bodies - Open Applications Group is supporting - RosettaNet, ANSI X12 and others are interested The UDEF teams in the AIA electronic enterprise groups have been working on several key items required to enable semantic equivalency across disciplines, industries and standards: - Basic data element concept trees, based on ISO 11179 object and property words: These have been created, maintained and communicated by Ron Schuldt. The trees need to be built out to reflect concepts within many disparate industries. We are now requesting participants and subject matter experts to help build out the content in the property and object word trees. The trees are viewable on the www.udef.org web site under Specs and Docs (UDEF Master Trees and Spec) - Registry/repository architecture use cases for semantic trees: The UDEF trees, or any open and globally available semantic identification framework will require a structure that supports multiple views, searching, expansion, transformation and various activities as yet unforeseen. The primary portion of this (the permanent home of the tree documents) is not a UDDI or ebXML registry/repository. The repository that would store standard-specific, software vendor-specific, or industry- specific ontologies and taxonomies labeled with UDEF IDs could be enabled via UDDI / WSDL or ebXML reg/rep technologies. A set of use cases, and some diagrams are viewable on the www.udef.org web site under Specs and Docs (UDEF Registry Use Cases). - Structure for creating XML versions of the trees to be stored in the above referenced reg/rep, searchable and capable of message based communication between nodes or other architectures. We are currently conceptualizing this as RDF, and addressed via namespace addressing for maintainability at the site of the matrix owner or perhaps hosting locations. Details on the initial efforts can be viewed at www.udef.org under Specs and Docs (Current UDEF XML Architectural Design PPT and zip file), provided by Kevin Mitchell and Al Gough at AMS http://www.ams.com. - A number of white papers, presentations for 101 and 201 tutorials and other materials, provided by Ron Schuldt, are available at www.udef.org under Education. - The AIA and EIDX groups have together collaborated on a Web Services based proof of concept. This POC provides two purchase order documents (OAGIS and xCBL) that are labeled with UDEF IDs in the document instances (not considered to be good form as a production ready structure for the placement of the semantic IDs) for the purpose of a UDEF Compare Report. This can be referred to in the WSDL spec for the service. The Compare Report application service can be viewed at www.udef.org under Proofs/Pilots. There is a detailed presentation that describes the effort, the approach and the architecture (Linux, AXIS and a servlet / web service for independent reference capability) with hosting generously provided by Envision http://www.envision.com, the core servlet provided by Jacek Ambroziak (jra@ambrosoft.com) at Ambrosoft http://www.ambrosoft.com, and the Web Services SOAP consumer and service created by Sudhir Suvva (sudhir.suvva@envision.com) under Ron Schuldt's and my direction. Planned efforts for Q3 2003 through Q2 2004: - I will be constructing Usage Scenarios for submission to available W3C groups such as Web Services Architecture and Web Services Internationalization. Ideas and assistance are welcome. - Implementation and Delivery Roadmap for standards bodies, software vendors, IT shops to collaborate on implementation approaches and pilots, specs, etc. - Selection of the .org entity to serve as steward, host and initial node for the registry/repository, facilitator of the management board (envisioned as a cross industry domain expert board) and the primary educational / certification authority. - Production launch of the above mentioned .org hosted registry / repository and related build out of the trees to reflect domain knowledge. "Schuldt, Ron L" <ron.l.schuldt@lmco.com> wrote: Date: Mon, 18 Aug 2003 08:56:30 -0600 From: "Schuldt, Ron L" Subject: Presentation at UDEF Conference To: hans.gerwitz@envision.com, angela.toppins@envision.com CC: "Decapua, David P" , "Bryant, William" , "Pallone, Michael J" , "Abt, Linda" , "Myers, Bradley B" , "Fadel, Chaker" , thomas.warner@boeing.com, djohnson@aia-aerospace.org, william@aia-aerospace.org, cookst@mail.northgrum.com, d6.smith@ngc.com, darcy.h.smith@boeing.com, johnchardin@yahoo.com, william.french@earthlink.net, golsen@contivo.com Attached is the presentation I will be making Tuesday at the National UDEF Conference in St. Louis hosted by Envision. See http://www.envision.com/feature/ for details. On Friday last week at the CompTIA/EIDX Conference, Oracle, SeeBeyond, IBM, GXS, DataDirect Technologies and Unicorn Solutions became interested in the UDEF. Since there is no known solution to the semantics integration problem (according to the experts during a panel discussion - IBM, SeeBeyond, DataDirect and Unicorn), they all wanted to learn more about the UDEF. Bill French, President of EIDX, was moderator of the panel on Web Services and encouraged them to learn more about the UDEF based Web Service that was demonstrated earlier in the week. I pointed them to the UDEF.ORG Web site. Based on a demonstration of Oracle's Web Service development tool, EIDX wants to pursue the possibility of using the Oracle tool to further refine the UDEF based Web Service. <> Ronald L. Schuldt Senior Staff Systems Architect Lockheed Martin Enterprise Information Systems 11757 W. Ken Caryl Ave. #F521 M P DC5694 Littleton, CO 80127 303-977-1414 ron.l.schuldt@lmco.com > ATTACHMENT part 2 application/vnd.ms-powerpoint name=UDEF-Convergence-of-Standards.ppt ********* john hardin webservices_and_integrations -- businessmodels::analyst::architect::manager http://www.geocities.com/johnchardin/ http://www.udef.org - subscribe to udef.builders-subscribe@topica.com > ATTACHMENT part 2 application/msword name=SemWeb-Cyberspace-Gnowsis.doc ********* john hardin webservices_and_integrations -- businessmodels::analyst::architect::manager http://www.geocities.com/johnchardin/ http://www.udef.org - subscribe to udef.builders-subscribe@topica.com ********* john hardin webservices_and_integrations -- businessmodels::analyst::architect::manager http://www.geocities.com/johnchardin/ http://www.udef.org - subscribe to udef.builders-subscribe@topica.com - To post to the discussion list, send an email to udef.builders@topica.com. - Archive found at http://www.topica.com/lists/udef.builders/read - To subscribe, send an email to udef.builders-subscribe@topica.com. - To unsubscribe, send an email to udef.builders-unsubscribe@topica.com. --^---------------------------------------------------------------- This email was sent to: www-rdf-logic@w3.org EASY UNSUBSCRIBE click here: http://topica.com/u/?a2i6DC.a6NeOr.d3d3LXJk Or send an email to: udef.builders-unsubscribe@topica.com TOPICA - Start your own email discussion group. FREE! http://www.topica.com/partner/tag02/create/index2.html --^----------------------------------------------------------------
Received on Monday, 15 September 2003 19:44:58 UTC