Put the digital into digital publishing

• Compose personalized documents for the reader. For example, once could have a text base on statistical methods, where each topic is treated at several levels (such as for readers who know calculus and those who do not) and where each method has examples from several domains. Given a reader’s background, subject domain, and learning goal, the system would then sequence of topics that build on each other (using prerequisite relationships and didactic relationships among topics). Select examples from the reader’s subject domain and voila there is a personalized statistics textbook. More generally this requires sophisticated algorithms for document structuring that are driven by an ontology of topic relationships