Many documents used in teaching and learning are composite: an exam paper contains questions, the essence of a reading list is a collection of citations, and a syllabus lists course components.

The "document" model, inherited from a print-oriented era, presents challenges to associative, non-hierarchical models of representation common in modern Web systems: "mashing up" becomes difficult when the level of granularity available is too large. For example, we may wish to navigate between exam questions on similar subjects for multiple years, or to the associated syllabus entry, lecture notes, or reading list entries: a task which currently involves a fair amount of "looking up" in multiple composite documents.This project will apply techniques to allow the production of "mashable" granular components from composite documents. The project will begin with exam papers, highly formulaic in layout, where such techniques can be easily applied for a substantial gain. Following an evaluation of the outcomes of applying this process for exam papers, attention will turn to related documents such as syllabuses and – particularly -- reading lists, hoping to build on earlier Arcadia projects in this area.