Discussion on the Discuss-DOI mailing list has focused on five topics over the past several months. They are:

1. Local versus global resolution of the Digital Object Identifier in the routing of queries on a data object
2. Interoperability standards
3. Event notification protocols
4. Intellectual property management
5. “Dumb” versus “Intelligent” DOIs

The topics tend to blend into one another as do the threads of the discussions about them. Let’s examine them in order.

1. The nub of this debate centers around the routing of queries about an item. Initial discussion dwelt on how such transactions were to be managed. Would the browser in a particular system be able to identify manifestations of the object in its local system or would all such queries have to be sent to a central or global authority which would identify various instances of the item, much like holdings data in a MARC record, and route the query to the nearest available one? An alternative approach proposed that standardized syntax and/or semantics within the DOI would enable the local system to route the query to the correct location first rather than to a centralized authority. If the item wasn’t found locally, the query would be shifted to the next higher (or larger) system and so on until the item was found.

   The debate outlined above illustrates concepts encountered in a number of other forums on metadata: interoperability, nesting, and standardization. The first approach operates on the interoperability aspect of the DOI. Given that the syntax and semantics of the DOI falls within proscribed limits (standardization), the local browser should be able to parse the query and locate the item, if it’s available, without outside intervention. The latter approach relies on the concept of system hierarchy. Standardization within the system occurs because each system fits within the larger one that encompasses it — the Ukrainian Doll model operating on a system level basis. Like buckets or containers containing different informational objects can be passed around because their containers adhere to a certain structure that the system can process, “understand” and locate.

2. The issue of interoperability standards arises out of the discussion above. The question of standards devolves into a debate concerning the merits of consortia-based standards versus those imposed by a public (read central)-authority. Another slant on this issue posits it in terms of the merits of proprietary technology versus those of standards. A further expression of the dichotomy of thought on this topic focuses on the abilities of local cache servers to route queries to local versions of the data object in question. One discussant mentioned the possibility of instilling “preferences” in the cache, but Leslie Daigle stated that local browsers currently are not sufficiently configurable to handle that kind of routing and/or filtering. Currently, all browsers in a particular environment treat Uniform Resource Names (URNs) the same way. He suggested employing proxy level settings in the local environment whereby items with DOIs would be done locally and items with real names would be handled on a global level. David Ades points out that the experience of major archival libraries in this area indicates that the increasing complexity in the development of such systems eliminates the practicality of maintaining copies of an object at a local level. It would seem that such systems hasten the consolidation and responsibility of the maintaining of holdings of such materials to a few major holdings libraries to which other libraries would refer their queries.

3. One method to determine the veracity of the implications mentioned above comes out of the notion of event notification protocols which allow the automated tracking of data flows. Such protocols would permit report updates and the collection of data to determine where data object queries were being routed. This proposal has the backing of both the IETF and the W3C.

4. The issue of intellectual property management drives and informs much of the technological discussion instanced above. The concept of the multiple resolution model enables publishers and other information providers the means to use one identifier which resolves to different versions of the object, whether that is the page of metadata, the object itself, or another manifestation of the object in a different medium. In this conception of metadata and the DOI, the functionality of the metadata falls out with the description of the purpose for which the particular DOI is employed. Adhering to the ICI format in the DOI suffix would enable the particular browser to identify the type of material engendered by the request with further extensions permitting the identification of individual instantiations of the data object at the local level (Much like locating a copy of an item in a particular library’s holdings in its OPAC).

5. This brings us to our last topic, “dumb” vs. “intelligent” DOIs. One discussant points out there already are “as many numbering systems as there are poppies in the field” which exacerbates the problem of resolving queries to the correct object. The current trend, backed by the NISO syntax group, favors the employment of dumb DOIs which refer to nothing other than the object itself in their role as a unique identifier. In this capacity as very structured (well-formed) metadata, they promise to work well with XML and other RDF initiatives. This approach tends to move the question of routing structure toward the container concept mentioned earlier. So long as the container type is properly identified, the objects within them can differ significantly within specified parameters.

   Work towards the solution mentioned above continues. A second draft summary of this work in progress is located at: http://dx.doi.org/10/1000/131. Norm Paskin who has been a prime motivator and developer of this paper stresses that it is not a consensus paper. He encourages discussion and comment at this web site. A third version with specific recommendations for a core DOI metadata element set will be published in mid-December. This paper promises to build upon the European Community’s Interoperability of Data in E-Commerce Systems (INDECS) work on structured metadata and its likely accord with Dublin Core to define a metadata standard for DOIs which will have the support of other metadata initiatives.