There's No Business, Like XOBIS:Background Material
XOBIS is a direct outcome of Lane Medical Library's earlier work, part of the ongoing Medlane Project (6, 28). In January of this year, an initial sketch of the unnamed schema was presented in a keynote address for a regional medical library meeting (29).It used our current structure, but had differing element names and interpretations. Lane's interest in XML is very pragmatic; we want to convert bibliographic and authority data into a Web-friendly format in order to integrate it with our other Web resources to facilitate Web-based initiatives. An open approach can help prevent important library resources from becoming marginalized, provide improved access and management of all our resources, and contribute strategically to the viability of libraries in an increasingly competitive milieu.
XOBIS represents one realization of the idea of organic bibliography, first posited in print in 2000, although brewing much earlier (30).In 1999, the Medlane Project produced the XMLMARC software, demonstrating the feasibility of flexibly converting MARC data to XML.However, it used an ad hoc DTD developed in about six weeks by two people and required more technical dexterity than desirable. While this represented only a cursory attempt to gather together various data elements dispersed in MARC, it helped bring into focus the problems of identifying elements for a more elegant representation of library information.
In contrast, Version 1.0a (alpha) of the XOBIS schema has taken over 6 months of part-time effort by roughly ten people to date, with work being delayed for a full year while implementing a new integrated library system at Lane (31).The RELAX NG schema, created by James Clark and Makoto Murata, was chosen primarily due to its simplicity in comparison with the XML Schema language, used by LC (32-34).
XMLMARC itself separates content from functionality by segregating software functions from transformational definitions recorded in a separate XML mapping document.In conjunction with the release of XOBIS, we are releasing Version 2.0 of XMLMARC (35).XMLMARC has been almost completely rewritten with changes to the mapping language, usability of the product, and underlying MARC application programming interface (API). Kevin Clarke recaps the program's design and functionality in recent publications (36-37). Default maps for "vanilla" MARC and several variations are envisioned. Custom mapping of any local practice and/or extensions to MARC could take advantage of added features that the XOBIS structure affords.Of course, the XMLMARC software will continue to allow for the conversion of MARC to XML that can be validated by any DTD or schema as long as users are willing to develop the necessary mapping document. Due to the fundamental reorganization of data in XOBIS, we do not believe it is possible to provide lossless mapping from XOBIS back to MARC.
A note about methodology: While a more consultative approach might have been engineered, committee efforts are not particularly noted for speed or fostering creative solutions.We believe that a smaller, highly focused collaboration was needed initially to achieve the desired degree of synthesis and structural fidelity. In the spirit of Raymond's "The Cathedral and the Bazaar", we hope that XOBIS, following our necessarily restrictive origination phase, can be refined and further developed (or even rewritten at least once) using the open-source model (38).Medlane Project staff, based in a medium-sized library (large for a medical library), considered as broad a range of data as practicality permitted. We benefited from varying degrees of experience and from sharing a wide range of backgrounds. This was supplemented by conversations with invited guests regarding various aspects of the design. It is interesting to consider that medium-sized libraries make good testbeds; they are large enough to exhibit most complexities realistically, yet small enough to not become overwhelmed with scale issues.XOBIS is by no means complete or comprehensive.After testing the initial alpha version, we anticipate releasing a beta revision. Admittedly, more could be done to address bibliographic issues, which may not have surfaced, or which were crowded out during our deliberations.At the very least, we hope XOBIS will serve as a springboard for discussion and development of a robust schema to take advantage of the many previously extolled benefits of XML (39). For those interested, we invite participation in the new XOBIS listserv (40). For XML in libraries in general, we suggest the XML4Lib listserv (41).
Notes about conventions used: