Tag Hierarchies

Background

Recently I have been interested in trying to create hierarchical taxonomiesfrom flat tag data. Tagging systems like del.icio.us, Flickr, and CiteULike tend to have (relatively) flattags. This means that while one can easily browse by a tag, like photography, one cannot aseasily see tags which are more or less broad than that tag. It is alsodifficult to get a broad overview of what tags exist in these sorts ofsystems as a result, aside from frequency based displays like tag clouds.

Some commentators have suggestedthat ontology is overrated, even irrelevant. That there is no hierarchy inideas, only links:

This may be overstating the point a little bit. While often manyhierarchies can be created for any given set of data, hierarchies areindisputably useful for a major type of information retrieval task:browsing. When we do not know exactly what we are looking for, it ismuch easier to be able to broaden and narrow our area of interest than toperform some sort of random walk from idea to idea. The top few categoriesof a traditional hierarchy give us a much better idea of the contents of amedia collection than thousands of individual tags, even if these tags areranked by their frequency in the collection.

Tagging systems are excellent at the task that they were designedfor---allowing a large, disparate group of users to collaboratively labelmassive, dynamic information systems like the web, media collections ofmillions of images, and so on. We are working to make these systems betterby automating production of hierarchical taxonomies that describe the datafrom the raw flat tags generated by users.

I've found some interesting features of tagging datasets from del.icio.us and CiteULike which have in turn suggestedreasonably good ways to create hierarchies. An example hierarchy generatedusing some of these methods from del.icio.us is here: mgfgsm-hierarchy.

Papers

Title:	Collaborative Creation of Communal Hierarchical Taxonomies in SocialTagging Systems
Authors:	Paul Heymann and Hector Garcia-Molina
Type:	Preliminary Technical Report
Accessible:	(info)(ps)(pdf)
Description:	This paper describes a simple algorithm for constructinghierarchies in social tagging systems that usually works reasonably well.The main contribution is a notion of generality in social tagging systemsbased on centrality in a similarity graph.