Tag Archives: taxonomies

Tag, You’re It!

Participatory web technologies in libraries are empowering users to engage with each other socially through reviewing, tagging, and rating content to potentially expose it to thousands of people who might otherwise never have seen it. Arguably, this is a very good thing.

Users with specific domain knowledge can provide a more detailed, comprehensive description of a resource than a harried librarian may be able to. Devoting their time and expert knowledge to enrich a record adds immeasurable value to the system as a whole. Furthermore, emulating the robust recommender systems of web giants such as Amazon can be provide new, meaningful pathways to content discovery.

Third-party services such as LibraryThing and GoodReads already provide many of these functions, and are fast gaining a great deal of traction, making it possible for users to participate in  catalogues—but these services still require external layers of software to function.

The problem with embracing this unchecked enthusiasm for social data lies in placing increased responsibility on unaccountable amateurs. Freely chosen words lack the structure or consistency demanded by cataloguing, and social data has a tendency towards mislabeling important data and misrepresenting contentious issues. Adversarial tagging or deliberate vandalism pollutes records, and personal tagging with no thought towards the overall context dilutes the usefulness of the system.

Existing cataloguing practices are overseen by the skilled work of experts, whereas the unmoderated nature of current social indexing platforms is lawless by comparison. Successfully integrating these participatory elements must be done with the utmost care.

A Pattern Language

I cannot escape the idea that resource description is an exercise in abstraction: thinking about huge sums of information and knowledge in ways that can be summarised, organised, and represented.

As an exercise, it asks me to consider fundamentally philosophical, ontological questions about the state of being. In my mind, ontology refers to the old philosophical questions of ‘what kind of things exist’, whereas the librarian, programmer, or taxonomist grapples with the question of ‘what kind of things should we capture and represent?’.

Taking the philosophical argument one step further, it is worth considering the work of empirical philosopher Ludwig Wittgenstein. Wittgenstein posed many questions about epistemology—the theory of how knowledge is acquired and represented—and argued that language permeates and defines reality. Words are tools, he argues, and instruments that provide the necessary, contextual clues to provide clarity to understanding. Wittgenstein posited that without agreed-upon ordinary language, understanding becomes frictionless ice that fails to describe or impart any knowledge.

Structured cataloging systems explicitly demand language be stripped down to its functional, tool-like purpose. Proper categorisation is the key to facing the exponential growth of texts and media, and consistent, structured metadata means we can find the most relevant documents in the least possible time.

Legacy systems for representing bibliographic information like ISBD are undeniably powerful, but were developed for a context that no longer exists. Bootstrapping successive iterations of new categories, descriptors, and data onto a system that was built in an era of card catalogs has lead to bloated, indecipherably specific codes that can only be parsed by machines and seasoned operators. The drive towards an inclusive, semantic approach to information systems has brought ordinary language to the fore of systems design, and the archaic codification of bibliographic records in ISBD is jarringly outdated.

The semantic approach to cataloging demands no less rigour in creating authoritative, consistent, and highly-descriptive records, but great strides are being made in the evolution of information standards such as RDA. RDA not only parses information in ordinary language, but adds relational layers of metadata that situate records in terms of context, not just content.

I’ll finish these musings with a quote from architect Christopher Alexander, who in discussing construction for his 1974 book Notes on the Synthesis of Form described language in a way that prophetically resonates with the future of semantic data we now face:

“We are searching for some kind of harmony between two intangibles: a form which we have not yet designed and a context which we cannot properly describe.”

(Alexander, 1974, p.26)