Presentation on theme: "Modelling web resources Ketil Albertsen, Paradigma project National Library of Norway."— Presentation transcript:
Modelling web resources Ketil Albertsen, Paradigma project National Library of Norway
Aggregates Documents (in a bibliographic sense) spread over multiple web pages Web pages consisting of multiple visual/sound components, or traditional multipart documents Web pages requiring a style sheet, font file etc. for presentation (not directly visible to user) Snapshots of a web page varying with time, the web version of integrating resources.
Aggregates Web documents available in multiple formats: HTML, PDF, DOC, … Web pages providing information in alternative languages or adapted to several web browsers Blogs and similar pages, the web version of incrementally issued continuing resources Expressions based on multiple creative efforts: A song (melody + text), rewritings, …
Claim: Implicit aggregates, based on objects having some common attribute value, are not sufficient Inspecting candidate objects for common attribute values is costly, even when using indexes. Access to attributes may require document interpretation, requiring knowledge of the format. An aggregate may span only a subset of objects having a given attribute value. Many aggregates are not based on stored, common object attributes. Aggregates frequently need an ID and possibly a bibliographic description.
Solution: Aggregate objects, boxes General container objects holding (references to) other objects. Boxes can be identified and described as composite objects. The model defines classes of boxes, each class identifying a set of structural properties. The application/user may identify multiple applications of each box class.
Box classes (suggested) Component: Ordered, all components presented Variant: Unordered, one of several. Synchronous: Parallel presentation in real time. Update: Later components replace earlier ones, integrating CR Extension: Later components are added to earlier ones, successively issued CR. Based On: One component is a further development of one or more other components. Fragment Of: One component is part of another one.
Box class properties Cardinality, i.e. 1:n, m:n, 1:1 Existence, i.e. will removal of one component imply removal of another Dependency, e.g. a component depends on a whole Recursion, e.g. a chapter may contain subchapters Ordering, usually in time or space Reference, does a dependent component have an independent value or is it integrated with the other object?
Other Paradigma extensions Agent objects, for modeling real world objects that cannot themselves be digitized. Point/fragment reference objects (further discussed tomorrow)