|Full text PDF:||http://hdl.handle.net/2160/13779|
Tagging is an ever-growing feature of online systems. As more and more content is tagged by users, the resulting “folksonomies” grow and can become unwieldy. The design of tagging systems must take into account how the resulting networks of tags are composed, and what motivated the taggers, in order to best use those tags as an aid toward search on the system. A literature review was carried out on the topics of folksonomies in general, how they compare with more formal ontologies and how folksonomies can be improved. Studies categorising folksonomy tags were analysed, with particular attention paid to those studies using the resulting categorisation information as a means to infer tagger motivation. A specific strand of the literature review focussed on studies of book-tagging systems. The aim of this study was to take a particular tagging system, the book website LibraryThing, and analyse the tags on fifty sample books. Long tail tags with a frequency of 2 or less, were ignored for reasons detailed in the research methodology, leaving a total of 13,358 tags to be viewed and categorised. The tag frequency distribution was shown to demonstrate the Zipfian power law. The tag categorisation model indicated that booksonomy taggers generally tag within the categories of “genre/style”, “subject” and “personal task-based”. Users are motivated mainly by their own personal organisational needs, but also by some social impulses towards the other users of the site. As a further component of the study, tags on two specific book genres (non-fiction and young adult) were analysed separately. Patterns such as higher-than-average tagging in the “target reader” category on young adult books became apparent. The original research and datasets generated for this study provide further source material as a contribution to the evolving discussion on folksonomies in general, and on booksonomies in particular.