A real highlight of this year’s Online Information Conference was the session this morning on Taxonomies and Metadata (day 3, track 2, session 1).
Joseph Busch from Taxonomy Strategies talked about Benchmarking Taxonomies in his session keynote. He described some very practical qualitative and quantitative methods for evaluating the usefulness and usability of a taxonomy which he and his colleagues have put into practice in a number of settings both within corporate intranets and on corporate web sites. The take home message was the benefit of adopting a “evaluate and improve” approach to the management of taxonomies. For more information see the presentation Benchmarking Your Search Function: A Maturity Model.
Jayne Dutra from the NASA Jet Propulsion Lab at Caltech presented a talk entitled Pushing the envelope at NASA: developing integrated taxonomies for a tiered information architecture. Jayne’s talk was packed with information about the economics and practicalities of actually implementing an infrastructure for managing information within the JPL, and an extremely valuable resource for anybody struggling with planning the concrete steps required to ultimately exploit the potential benefits of taxonomies and metadata.
Tom Reamy of the KAPS Group gave a talk on Folksonomies and complexity theory: evolving information structures. I’ve been fascinated by complexity theory ever since I was pointed at a book called “Complexity” by M. Mitchell Waldrop – I immediately felt the underlying ideas were enormously useful, but have been struggling to understand how to apply it in the fields of metadata and search, beyond the obvious ideas regarding social networks and link topology. Tom presented a project where the basic principles of complexity theory – local rules, local interaction, feedback – have been applied to design a system which leads to the emergence of useful structure, starting from a unstructured folksonomy. The hope is that, in so doing, the drawbacks of folksonomies – polysemy, synonymy, no structure, jargon, no quality control … – can be overcome, without losing any of the advantages – simple, low cost, open ended, relevance etc.