2003-08-29 Pluggable user models for adaptive hypermedia in education Abstract from http://www.ht03.org/papers/ "Most adaptive hypermedia systems used in education implement a single user model Ð inevitably originally designed for a specific set of circumstances. In this paper we describe an architecture that makes use of XML pipelines to facilitates the implementation of different user models" * education on www is all too often a "distributed photocopier machine" -> adaptive hypermedia can help * user models are imperfect -> user models should be plugins -> can be done as filters in the XML pipeline (series of transformations of XML) [i'm not sure this works. the user model has to hold the user map and all that kind of stuff, but the implicature in a conversation can *utterly* change the speech act that moves you round the territory. it's not just a perturbation, which is essentially what a pipeline models. maybe that doesn't matter here because it's very much a trail - the lesson - and not so much navigation.] system is called "WHURLE" divided into chunks: atomic unit of content, conceptually self-contained (defined in WCML), written by subject specialists -- chunks are tagged with the user model a lesson plan creates a docuverse XInclude used to retrieve chunks as required. hypermedia pathway through available chunks. created by teachers. pipeline: visualisation of the content, link overlay (navigational are automatic, and authored links), skin overlay future work: - user models that aren't *just* about ability level (beginner, learner, expert) in a particular lesson/domain there are 60 different styles of learning preferences, and they want to model some of these, the ones appropriate to learning in nottingham [oooh, i want a reference for that 60 figure] didn't hear the answer about learning styles very well, but... felder & liverman models of learning style categories/models? there are various types, and you can match chunks to different model categories and adapt based on that.