Good Tags – Bad Tags: Social Tagging in Knowledge Organization
Monday, February 25th, 2008
Tagging as a lightweight, more participatory approach to resource metadata has become a hype in a wide range of disciplines, among them computer science, information management, and psychology. Last week’s workshop “Good Tags – Bad Tags: Social Tagging in Knowledge Organization”, well organized at Tübingen at the Knowledge Media Research Center, showed the different, but converging perspectives on tagging. Main success factors identified were:
- simplicity (and the resulting ease of use), although a “tagging competency” was still considered necessary
- enabling of participatory system design (i.e., tagging as appropriate metadata formalism for user generated content)
- reduction of cognitive load for the annotator
- flexibility and agility (not prescribing metadata structure, but allowing for emergence thereof)
The workshop also showed different understanding of “tags” and “tagging”. Johannes Busse (a colleague from ontoprise working in our project Im Wissensnetz made a good point with differentiating between linguistic tagging (where a tag represents a syntactic string) and semantic tagging (where a tag is another resource, be it a semantic concept, a person, another resource etc.). Also, some definitions of a tag annotation were presented, which stressed that a tag annotation is basically a triple (resource, tag, user) with several other attributes (e.g., timestamp, but also additional semantic qualifiers).
While tagging is a challenge for controlled vocabularies like library thesauri, it should be clear that controlled vocabularies and ontologies have their clear benefits in terms of disambiguation, multilinguality, levels of abstraction, and simple typos. So our contribution on ontology maturing, from simple tags to full-fledged ontologies actually showed the way: leverage on the simplicity of tagging for broadening the group of users contributing to the metadata generating, while retaining the possibility of higher-quality metadata. Interesting lightweight developments in that respect was the approach on Semantic Weblog by Benjamin Birkenhage.
