The evolution of web technologies and social trends fostered a shift from traditional enterprise databases to web services and online data. While making data more readily available to users, this evolution also raises additional security concerns regarding the privacy of users and more generally the disclosure of sensitive information. The implementation of appropriate access control models is one of the approaches to mitigate the threat. We investigate an access control model based on (non-materialized) xml views, as presented among others in~Fan et al. The simplicity of such views, and in particular the absence of arithmetic features and restructuring, facilitates their modelization with tree alignments. Our objective is therefore to investigate how to manipulate efficiently such views, using formal methods, and especially query rewriting and tree automata. Our research follows essentially three directions: we first develop new algorithms to assess the expressivity of views, in terms of determinacy, query rewriting and certain answers. We show that those problems, although undecidable in our most general setting, can be decided under reasonable restrictions. Then we address the problem of handling updates in the security view framework. And last, we investigate the classical issues raised by schemata, focusing on the specific ''determinism'' requirements of DTDs and XML Schemata. In particular, we survey some techniques to approximate the set of all possible view documents with a DTD, and we provide new algorithms to check if the content models of a DTD are deterministic.
