| Abstract: |
The paper proposes a framework that explores document structures, metadata, and context to adapt mathematical documents on three layers - the content, structure, and presentation layer. On the presentation layer mathematical expressions are rendered according to a user context (defining, e.g., the users’ preferred convention, language or discipline). On the content layer, appropriate paragraphs of texts are selected that best suit the users’ individual information need as well as narrative and semantic context. On the structure layer, these content parts are arranged appropriately, while taking semantic and narrative interdependencies between parts
into account. Technically, the framework models documents as dependency graphs, where nodes correspond to any addressable part of a document and edges denote narrative and semantic dependencies. To allow for the adaptation on all three document layers, these dependency graphs are enriched with presentational, content, and structure variants. Finally, the enriched graphs are processed to generate a user-specific document. Users can guide this adaptation process by prioritising their individual constraints, narrative or semantic context. |