Stijn Dekeyser

Path locks for XML document collaboration

The hierarchical and semistructured nature of XML data can cause complicated update-behavior. The updates are not limited to entire document trees, but can involve subtrees and even individual elements. These document parts correspond to, e.g. sections in text documents or sub-diagrams in vector graphics files. Providing suitable locking mechanisms for semi-structured data can significantly improve collaboration systems that store their data as XML documents. We show that concurrency control mechanisms in CVS, relational, and object oriented database systems are inadequate for collaboration systems based on semistructured data. We therefore propose a new locking scheme of fine granularity based on path locks. We also show that our proposed mechanism avoids conflicts by ensuring serializability, supports both top-down and bottom-up query evaluation, and is relatively efficient.

A Transaction Model for XML Databases

The hierarchical and semistructured nature of XML data may cause complicated update behavior. Updates should not be limited to entire document trees, but should ideally involve subtrees and even individual elements. Providing a suitable scheduling algorithm for semistructured data can significantly improve collaboration systems that store their data—e.g., word processing documents or vector graphics—as XML documents. In this paper we show that concurrency control mechanisms in CVS, relational, and object-oriented database systems are inadequate for collaborative systems based on semistructured data. We therefore propose two new locking schemes based on path locks which are tightly coupled to the document instance. We also introduce two scheduling algorithms that can both be used with any of the two proposed path lock schemes. We prove that both schedulers guarantee serializability, and show that the conflict rules are necessary.

A commit scheduler for XML databases

The hierarchical and semistructured nature of XML data may cause complicated update-behavior. Updates should not be limited to entire document trees, but should ideally involve subtrees and even individual elements. Providing a suitable scheduling algorithm for semistructured data can significantly improve collaboration systems that store their data -- e.g. word processing documents or vector graphics-- as XML documents. In this paper we improve upon earlier work (see [5]) which presented two equivalent concurrency control mechanisms based on Path Locks. In contrast to the earlier work, we now provide details regarding the workings of a commit scheduler for XML databases which uses the path lock conflict rules. We also give a comprehensive proof of serializability which enhances and clarifies the ideas in our previous work.

DISTRO: a system for detecting global outliers from distributed data streams with privacy protection

In this demo proposal, we present a new system, called DISTRO (a.k.a DIstributed STReam Outlier Detector), for detecting outliers from distributed data streams. DISTRO is able to effectively identify outliers from distributed data streams that are consistent with those generated by the centralized detection paradigm. DISTRO is also able to ensure high-level data privacy throughout the detection process. A number of optimization strategies are devised to further enhance its speed and communication performance. This proposal provides details on the motivation and technical challenges of detecting outliers from distributed data streams, presents an overview of DISTRO, and gives the plans for its system demonstration.

Multi Query Optimization Using Query Pack Trees

Query optimization has long been one of the prime research areas associated to the relational database model. Most of the research assumed that users would pose unrelated questions, at relatively large time intervals. Recently, however, applications were discovered where a database is issued a multitude of queries simultaneously, while the queries themselves are some how related to each other. One such application is data mining. The topic of the Ph.D. research presented in this paper is the study of formal systems that make multi query optimization -- independent of the database instance -- possible.

Five examples of web-services for illustrating requirements for security architecture

The tension caused by the need for expressive power when formulating security rules and the need to keep computational complexity low when undertaking the necessary access rule evaluations is a major challenge in the formulation of good security architecture. This paper provides five examples of security in web services. which illustrate this tension. These examples highlight the need for more expressiveness in the rules used to express policies in three cases, and in the other the fact that XACML appears to have nearly adequate expressiveness without undue complexity. Each example is expressed first informally, by describing a service which could concievably be provided in a web services architecture, then the example is also outlined using either XACML, first order logic or both.

Guaranteeing Correctness for Collaboration on Documents Using an Optimal Locking Protocol

Collaboration on documents has been supported for several decades through a variety of systems and tools; recently a renewed interest is apparent through the appearance of new collaborative editors and applications. Some distributed groupware systems are plug-ins for standalone word processors while others have a purely web-based existence. Most exemplars of the new breed of systems are based on Operational Transformations, although some are using traditional version management tools and still others utilize document-level locking techniques. All existing techniques have their drawbacks, creating opportunities for new methods. The authors present a novel collaborative technique for documents which is based on transactions, schedulers, conflicts, and locks. It is not meant to replace existing techniques; rather, it can be used in specific situations where a strict form of concurrency control is required. While the approach of presentation in this article is highly formal with an emphasis on proving desirable properties such as guaranteed correctness, the work is part of a project which aims to fully implement the technique.