The Complexity of Text-Preserving XML Transformations

• Main Authors: Timos Antonopoulos, Wim Martens, Frank Neven
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Published: 2011
• Subjects: Categories and Subject Descriptors H.2.3 [Database Management; Languages—Data manipulation languages (DML; F.4.3 [Mathematical Logic and Formal Languages; Formal Languages; D.2.4 [Software Engineering; Software/Program Verification; DML
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.302.4102; http://fox7.eu/wp-content/uploads/pods11a-antonopoulos.pdf

While XML is nowadays adopted as the de facto standard for data exchange, historically, its predecessor SGML was invented for describing electronic documents, i.e., markedup text. Actually, today there are still large volumes of such XML texts. We consider simple transformations which can change the internal structure of documents, that is, the mark-up, and can filter out parts of the text but do not disrupt the ordering of the words. Specifically, we focus on XML transformations where the transformed document is a subsequence of the input document when ignoring mark-up. We call the latter text-preserving XML transformations. We characterize such transformations as copy- and rearrangefree transductions. Furthermore, we study the problem of deciding whether a given XML transducer is text-preserving over a given tree language. We consider top-down transducers as well as the abstraction of XSLT called DTL. We show that deciding whether a transformation is text-preserving over an unranked regular tree language is in Ptime for topdown transducers, EXPtime-complete for DTL with XPath, and decidable for DTL with MSO patterns. Finally, we obtain that for every transducer in one of the above mentioned classes, the maximal subset of the input schema can be computed on which the transformation is text-preserving.