Taxonomies are nice but far too rigid. The colon classification system proposed by Ranganathan in 1933 was the first serious attempt to propose faceted classification as an alternative, but the technology for faceted search took a while to develop (check out some early technology in the space).

Fast forward to the present day, and you have systems like WorldCat or the Endeca-powered system at the Triangle Research Libraries Network. These systems leverage facets to facilitate exploratory search. There are still taxonomies in there, but less emphasis on achieve one static hierarchy to rule them all, and instead a focus on leveraging context dynamically to propose useful query refinement options.

Librarians have a standard taxonomy (Dewey decimal) to help them get started in the information retrieval. If all of the books were thrown into a huge pile, it would be a pain for the librarian to find the best book for the user. Unfortunately, a lot of the web has a limited taxonomy or reference to help some one do the exploratory search.

I haven’t yet read all of you exploratory search writings but agree with that approach. I’m interested in the problems you are approaching and how taxonomy (static or possibly dynamic) would help drive the faceted search and exploratory search.

Stefano, I realized after posting this that I might have offended librarians by suggesting that that technology will make them superfluous. I do think that technology should automate much of what consumes librarians’ time today. For example, I was reading about the Hutchins heuristic, and it’s clear that librarians should not be executing algorithms that could be implemented in software. That said, it will be a while until our expert systems are competitive with human experts. And here I speak of experts in information seeking–figuring out better ways to leverage collective content expertise is a topic in and of itself.

David, I feel your wife’s pain. I look at libraries like the Triangle Research Libraries Network as a starting point for how online catalogs should work. My colleagues and I at Endeca are working on even better for support for exploratory search. As I’ve blogged here before, I think exploratory search is where it’s at.

The context around the source of data is so valuable. Understanding of the sources of the data, their validity, how current the information is, and whether that data will answer the question is as important if not more important than the simple relevance ranking.

These problems will keep us busy for quite some time.

I believe there are two important ways great librarians can help. Unlike automated systems, human assistants can help the searcher refine his or her question. Arriving at the right question is not a simple matter of getting the right query syntax: it’s about deepening your understanding of what it is you really want to know. We can all get better at that!

In addition, librarians bring a lot of practical context to prioritizing results. Like so many of the other mediators who are being “disintermediated” by the Internet (editors, publishers, teachers, journalists, producers, and so forth), librarians used to help us understand which sources we could trust and which were most likely to help us answer our specific question. It is much more than a calculation of relevance. It depends as much on a deep understanding of our questions and the authors and institutions which produced the text as it depends on the text itself.

If I am right about this, and I suspect that you agree, Daniel, both people and technology will be working together on this problem for a long time. And librarians can expect to play a major role in the “re-intermediation” process.