Oct. 30: Ding, Golder and Huberman

Ying Ding's article, "A review of ontologies with the Semantic Web in view," discusses several important ontologies in relation to human-computer interaction. Ontologies "can be seen as metadata that explicitly represent the semantics of data in a machine-processable way" (377). A widely cited definition of an ontology comes from Gruber: "an ontology is a formal, explicit specification of a shared conceptualization" (378). What is important here for the information science community is an ontology's relationship to the Semantic Web, which allows for machine-readable information exchange. Ding lists several important ontologies, ontology languages, and ontology tools. Each community may have its own specialized ontology. For example, the business community uses Enterprise Ontology, which highlights terms related to processes and planning, the structure of organizations, high level planning, and marketing and selling goods and services (379). Ontology languages "are either logic-based (frame logic), or web-based (RDF, XML, HTML)" (379). Continuing with our business ontology example, Enterprise toolsets "are implemented using an agent-based architecture to integrate off-the-shelf tools in a plug-and-play style" (380). Enterprise toolsets support the Enterprise Ontology discussed earlier. Ding also lists several ontology projects, including Enterprise, which is "aimed at providing a method and computer toolset which will help capture aspects of a business and analyse these to identify and compare options for meeting the business requirements" (381). Although much of Ding's article was very abstract to me, I understand the importance of ontologies with regard to the Semantic Web. Ontologies provide a set of standards which can support the interoperability of common tools and aid in their design.

On a lighter note, Scott Golder and Bernardo Huberman look at Del.icio.us, a popular site for bookmarking and tagging URLs. The authors discuss the difference between collaborative tagging, such as is the practice in Del.icio.us, and taxonomies, which are more hierarchical and exclusive. With collaborative tagging, individuals make the distinction as to what tag to apply to a certain bookmarked URL, which is influenced by the individual's level of expertise as well as social factors such as language and culture. Although collaborative tagging does present some problems, it also provides the "opportunity to learn from one another through sharing and organizing information" (201). Golder and Huberman looked at data from Del.icio.us to reveal patterns of use. They found that users initially prefer more general tags and that successive tags were more specific and/or personal in nature. Another important finding is that users often imitate other users and share knowledge in the network, meaning that they often choose tags that have been created by other users because they perceive them as being 'correct' when they may not know how to tag a particular URL. The authors assert that this factor may be a cause for the stabilization of tags to describe URLs. Interestingly, Del.icio.us in this way can be seen as a URL recommendation service "even without explicitly providing recommendations" (207).

I've never used Del.icio.us myself but, after reading Golder and Huberman's article, I am interested to see how it all works. As I was reading I was reminded a lot of Flickr, a photo storage service that allows its users to tag photos to be searched by other users. Flickr is, in my opinion, much more personal, or at least it can be, as it allows users to tag their own photos with personal names of friends, family, and even complete strangers. Of course, other more general tags can be and are used in Flickr. Searching through the millions of photos can provide hours of time-wasting fun!