_Ontological, Agent-based, On-the-Fly Cataloging_ 

     Recently, I posted queries requesting consideration of viewing _Library Subject and Organizational Systems as Ontologies _ as well as two related to the application of Intelligent Software Agents for Cooperative/Collaborative Collection Development. Concurrently, I also have been reviewing various Metadata initiatives (MARC, Dublin Core, EAD, GILS, MCF, SHOE, TEI, etc. [See IFLA "Metadata Resources" page at http://www.nlc-bnc.ca/ifla/II/metadata.htm] for mention in my presentation at the LITA SIG Authority Control in the Online Ennvironment program at the ALA annual meeting on June 28, 1998 2-5:30 [The theme is "Metadata and Authority Control" and all are invited; no reservation are required {:->] 

     My superficial review of Ontologies within the context of MARC as well as my superficial review of Agents and their _possible_ application to Collection Development, and other library services, has led me to consider the possibility of using agents to 'assemble' Ontologies that would serve as a 'catalog' record for a given information resource (e.g., print or electronic monograph, print or electronic serial title,  print or electronic serial article, Web pages/sites, etc. In this view the Ontology would substitute for a Cataloging Record (e.g., in MARC) or for a metadata presentation description (e.g., Dublin Core). In essense the Ontology would become the Metadata and Presentation Format. My thinking about this approach has been inspired by Thomas R. Gruber, one of the exceptional minds of this world, who long ago (1992 {:->) created a _Bibliographic-Data Ontology_ [See: http://www.aic.nrl.navy.mil/~harrison/ontologies/bibliographic-data/ index.html ; for the full outline of his bibliographic ontology]. [Peter Weinstein's work at well with the Digital Libraries Initiative at Michigan also had a direct affect on my thinking. I highly-recommend a visit to his project and publications pages. His paper prepared for this months ACM Digital Libraries conference is well-worth The Read. His _Ontology-based Metadata_  and _UMDL Ontology Concept Descriptions_ are of particular note [Available At/From: http://www-personal.umich.edu/~peterw/Ontology/Beethoven/demo.html] 

     In considering Ontologies, I initially thought of representing standard library classification and subjects systems as Ontologies. Indeed Vickery in a recent issue of _The Journal of Information Science_ [Vickery, B.C. "Ontologies." Journal of Information Science 23, no. 4 (1997): 277-286] presents an _excellent_ summary and profile of Ontologies as well as explicit recognition of Library Classification and Subject Systems as Ontologies [Thanks B.C!][BTW: I wish to publicly acknowledge the direct influence on my initial thinking about Ontologies to Jane Greenberg, University of Pittsburgh [Go Pitt! {:->]  Her excellent article "Reference Structures: Stagnation, Progress, and Future Challenges" in the September 1997 issue of _Information Technology and Libraries_ 16(3) [Abstract available at: http://www.lita.org/ital/ital1603.htm#greenberg] explicitly details the Ontological Nature of Library Schemes and I consider it a Must-Read! Thanks, Jane!] 

      On the way to work  [I get some of my best ideas riding the bus!], it occurred to me that perhaps we need to re-think our current approach to 'Catalog Record' for Print and/or Electronic resources. Not only should we begin to consider the Ontology as the Record Format, but to think more 'dynamically' about the nature of the data or information that would be included in each Ontology Class that constitutes the Ontology Format [See Gruber's and Weinstein's outline above] Instead of having one and only one appropriate member within a given class, there could be many such members. For example, in Gruber's Bibliographic Ontology in the Author Class we would have more than one Author, or Title, or Publisher, etc. These members would exist as Class Clusters and ranked by some systen algorithm within the Bibliographic Ontology for a given resource (e.g., monograph, serial, etc.) Such clusters would not be created manually but would be categorized by Ontological Agents that would analyze and *class*ify the 'content' of an 'object' (e.g., electronic book) based upon the function of each individual ontological agent. For example, within the Cataloging Ontological Agent, there would be an Author Ontological Agent to identify the 'Author(s)' of a 'work', and a separate Subject Ontological Agent to identify'Author(s)' of a 'work', and a separate Subject Ontological Agent to identify the 'subject' of a 'work', etc. The Cataloging Ontological Agent would organize the results of the sub-agent activities and present to the user a Catalog Record organized by the established Ontological Outline [See Gruber, for example], or customized to user preference. 

      Whew! Whew! Whew! 

      In summary, this model seeks to reconceptualize the cataloging process and the presentation and navigation of catalog data/metadata! 

      For my continuing search for The Truth and the Meaning of Life [:-)], I' d appreciate any reactions or elaborations on this proposed model. As Always, Any and All ontributions, questions, concerns, comments, citations, or queries are Most Welcome! 

   Joy! 

/Gerry McKiernan Curator, CyberStacks(sm) Iowa State University Ames IA 50011 [log in to unmask] http://www.public.iastate.edu/~CYBERSTACKS/ 



"The Best Way to Predict the Future is to Use An Agent! {;->] With Possible Apologies to Peter Drucker 



  




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%