Validation of Application Semantics with XML Schema

Abstract

The eXtensible Markup Language (XML) has evolved to become the de facto industry standard for the sharing and integration of data. This is mainly due to the syntactic flexibility by which end users can produce XML documents. Unfortunately, this flexibility is an inhibitor when it comes to the validation of XML documents with respect to application semantics. XML schema languages such as document type definitions and the standard XML Schema provide rich constructs to impose many structural constraints, but have significant shortcomings when it comes to semantics. Keys, as proposed by XML Schema, are not naturally defined and show provably bad computational properties. The research literature has proposed alternative key languages which are reminiscent of keys from databases, do not require an XML Schema definition, and have provably good computational properties. So far, the validation of the alternative key languages has not been studied in the research literature. It is neither obvious how to specify the semantics of such keys and how efficient a potential validation of such semantics would be. The first main contribution of this thesis is to express keys, as proposed by Buneman et al. [10, 12], in XQuery. It was found that XQuery provides features that can express complex key semantics in a logically structured and simple way. It is stressed that the implementation is capable of expressing XML keys beyond the path languages proposed by Buneman et al. [10, 12]. An example is given of how to validate the XML keys as assertions in XML Schema 1.1, too. Note that assertions are a key feature of XML Schema 1.1, not available in version 1.0. The second contribution of this research is to analyse the time-complexity of the validation process. The findings show that the validation works efficiently for rather complex XML keys on modestly-sized documents. However, the price for our natural implementation of XML keys is that the validation does not appear to scale well to larger XML documents. For such documents, it would be useful to invest future research into more complex implementations that scale better. The results in this thesis contribute to an increase in the semantic validity of XML documents, fewer flaws in the integration and exchange of documents, better decision-making capabilities based on less invalid data, and lower costs for data cleaning.