| Managing multiversion documents is important for traditional applications, such as software configuration and cooperative work, and for a fast-growing array of web-based applications. Traditional document version control systems, e.g. RCS, which model documents as a sequence of lines of text and use the shortest edit script to represent version differences, can be inefficient and do not preserve the logical structure of the original document. We first propose a new storage strategy, called the Usefulness Based Copy Control (UBCC), which improves on RCS by assuring more efficient version reconstruction with a small storage overhead. The improvements provided by UBCC notwithstanding, edit-based representations suffer from limited generality and flexibility—and do not support queries well. To solve these problems, we then propose a reference-based version scheme that preserves the logical structure of the evolving document via the use of object references. By preserving the document structure among versions, the new scheme facilitates efficient query support and provides an effective representation of multiversion XML documents, both at the storage level and the transport level. In fact, for the storage level, we develop a usefulness-based management policy similar to the UBCC strategy, and for the transport level, we represent the history of a multiversion document as yet another XML document.; Finally, we explore another direction of modeling the versioned document structure—the SPaR scheme. The SPaR scheme quantifies the document structure by assigning a unique node number to each element. In addition to traditional queries, this approach is efficient for new XML-oriented queries, including searches by regular path expressions that are used frequently in XML query languages. |
