Presenting multi-language XML documents : an adaptive transformation and validation approach

XML addresses several HTML shortcomings, but its underdefined processing impedes the development of adequate generic presentation models for the Web. Such models must define the parsing, validation, transformation and rendering of multi-language XML documents, according to a variety of adaptation requirements. However, most existing approaches only define subsets of this functionality and do not follow the Web design principles. We hypothesise that generic document presentation can be achieved by utilising the presentation domain constraints and addressing the document processing problem as a whole. This thesis focuses on the document preprocessing domain and supports our hypothesis by proposing a preprocessing framework and the XMLPipe preprocessing model. Document preprocessing is the document presentation subset that only addresses parsing, validation and transformation. The preprocessing framework establishes the necessary preprocessing functionality and enables the evaluation of XMLPipe. XMLPipe utilises the presentation domain constraints to provide generic XML preprocessing. XMLPipe consists of an integration model, an adaptation model, a transformation model, a validation model and a binding model. The integration model utilises the presentation domain constraints to infer a multi-language document’s interpretation from the interpretation of its constructs. The adaptation model proposes an extensible representation of the adaptation requirements and a method to choose the optimal processing alternative among a set of independently developed specifications. The validation and transformation models use the integration model to validate and transform multi-language documents, according to a set of adaptation requirements and a distributed set of processing specifications. The binding model establishes a distribution of the processing specifications, which is adequate for processing an open set of independently developed XML languages. The XMLPipe document processing is demonstrated to be significantly more powerful than existing approaches and its evaluation illustrates its adequacy for the Web and the soundness of our hypothesis within the preprocessing domain. The preprocessing observations are extrapolated to confirm our hypothesis within the complete document processing domain.