- From: Mark Donoghue <mark@thirdstation.com>
- Date: Tue, 01 Mar 2005 14:15:29 -0500
- To: Frank Manola <fmanola@acm.org>
- Cc: semantic-web@w3.org
Hello all, What a serendipidous opportunity! I was just today going to work on sending this information out to selected members of the list but, since you asked ... Semantic Web and the Digital Object Identifier - Persistent, Unique, and Scalable Resources for Semantic Web Applications - The company I work for is Content Directions, Inc. Our principal product is based on the Digital Object Identifier which, in turn, is an application of the CNRI Handle System. That's the short, obscure version so I'll clarify. CNRI Handle System (http://www.handle.net) ======= The CNRI Handle System is loosely described as a next-generation DNS or, DNS on steroids. It was invented by Dr. Robert Kahn, one of the principal architects of the Internet, and is his answer to the scalability problems of current naming systems. The basic unit in the Handle System is the "Handle" which is a valid URI taking the form of: hdl:<something-something>/<something-something>. What handles offer over current naming schemes are: 1) Persistance. Handles don't change. The resources they point to may change but, the handles remain the same. 2) Global Uniqueness. Handles are guaranteed to be globally unique within the global handle namespace. 3) One-to-many relationships. A handle record provides a level of indirection for a web resource. The handle itself does not point to the location of a resource instead, a handle record contains a value of the type "URL" and whose value is the current location for the resource. If the location of a resource changes you just need to update the handle record and all requests for that resource will be directed to the new location. An especially useful property of handle records is the ability to attach arbitrary data to any handle and, it's very RDF-like. Basically, you say "This handle has a value of type so-and-so whose value is such-and-such." Digital Object Identifiers (http://www.doi.org) ======= The DOI Foundation website describes DOIs best so I'll quote directly here: "The Digital Object Identifier (DOI) is a system for identifying content objects in the digital environment. DOIs are names assigned to any entity for use on digital networks. They are used to provide current information, including where they (or information about them) can be found on the Internet. Information about a digital object may change over time, including where to find it, but its DOI will not change. "The DOI system provides a framework for persistent identification, managing intellectual content, managing metadata, linking customers with content suppliers, facilitating electronic commerce, and enabling automated management of media. DOIs can be used for any form of management of any data, whether commercial or non-commercial. " The DOI system uses the concept an Application Profile to describe the semantics governing the values associated with any particular DOI. A DOI is a handle so, DOIs are valid URI as well and take the form: doi:10.1570/m.donoghue The above is a DOI that identifies me, Mark Donoghue. The parts are broken-down thusly: doi: 10.1570 / m.donoghue ---- ------- ---- ---------- scheme prefix a slash local identifier This DOI unambiguously identifies me, contains information about me, and is controlled by me. Every entity wishing to register DOIs is issued a prefix by a registration agency (like Content Directions). They are then free to create and register DOIs, usually with the help of the registration agency (that would be us again). Content Directions, Inc. (http://www.contentdirections.com) ======= As I mentioned, Content Directions, Inc. (CDI) is a DOI registration agency. In addition to providing DOI registration and creation services we add value to the DOI with what we call MultiLinks. A MultiLink is a DOI value that is used to make an association between the DOI it's attached to and another web resource. We use the MultiLink values to create a menu of choices associated with the DOI. Many examples can be seen at http://doi.contentdirections.com/. A DOI MultiLink value takes the form of a key=value pair where the key is used to create a label and the value is usually a URI pointing to a related DOI or a web page containing related content. As an exampple of how DOIs (which are handles) are RDF-like: An RDF statement about Content Directions could be: <Content Directions> has a <website> which can be found at <http://www.contentdirections.com/> Now, Content Directions happens to have a DOI and it is doi:10.1220/1. So, to map the above statement to the DOI system, instead of the string 'Content Directions' we'd use doi:10.1220/1 and, instead of <website> we'd use <URL> so it would look like: <doi:10.1220/1> has a <URL> whose value is <http://www.contentdirections.com> For an example of MultiLinks we can turn to the DOI that references me: <doi:10.1570/m.donoghue> has a <MULTIRES> whose value is <Website=http://www.thirdstation.com/> To see what this statement looks like in real life point your browser to http://hdl.handle.net/. Once there, put my DOI (10.1570/m.donoghue) into the textbox labeled 'Handle'. Put a check in the box labeled "Don't Redirect to URLs," then submit the form. What you will see is the actual handle record for my DOI. The above RDF statement should be the first item in the list with an index of 1000. ======= And this brings me to my final point. HOORRAAAYYYY!!!! When you look at my DOI handle record you will also see a bunch of handle values that look like FOAF.* Those are what I've been working on for the past few weeks and are a demonstration of how Friend-of-a-Friend data can be mapped to DOIs and used by clients to create social networks, trust networks, and topic maps among other things (left as an exercise for the reader). In addition to FOAF data I've been toying with embedding Dublic Core and PICS into DOIs. I've also created DOIs from RSS feeds (see http://www.contentdirections.com/rss_demo/). PICS labels would be easy to add to DOIs for clients to use when filtering web content and the label is right there with the location of the content. An additional file containing the labels is not required. ** Here is what we need ** Partners. We need partners. I can play with this stuff all day but, in the end, we can't afford to chase down and create a market for every application of the DOI. We want to partner with persons/companies who wish to create pilots using the DOI system as the backend for their Semantic Web application. We would supply the access to, and expertise with the DOI system. You would work with your customers and us to implement the pilot. Paid pilots are preferable but, if the application is compelling and the customer is high-profile we are willing to negotiate. Our primary goal is to get the DOI into wide public circulation. It's my opinion that the Semantic Web is where we can have the greatest impact. I'm hoping you can help me prove that to be correct. Thanks, Mark /********************************************* * Mark Donoghue * Consulting Engineer * Content Directions, Inc. * * p: (845) 242-8501 * e: MDonoghue@ContentDirections.com * doi: http://dx.doi.org/10.1570/m.donoghue */ On Mar 1, 2005, at 9:41 AM, Frank Manola wrote:
Received on Tuesday, 1 March 2005 19:26:17 UTC