Semantic Modelling of User Interests based on Cross-Folksonomy Analysis
They ask, what use are these online profiles of interests? One application would be personalization. They observe that individual’s tag clouds often suggest what they are interested in. They gather and correlate tags from a user’s many different accounts (they find them by you typing in your home page url; this is generally associated with your account in many tagging systems), trying to regularize against between-account variations (e.g. when some tagging systems permit whitespace and others don’t). They then lookup through wordnet and wikipedia categories to translate from specific tags into general topics, and shove the resulting tags into your foaf file as a list of interests. The work seems interesting but incomplete: they identify a bunch of words that sound plausible, but they never checked whether they actually describe the interests of the users from which they were derived.