Efficient schemes for managing multiversion XML documents

Chien, S.-Y.; Tsotras, V.J.; Zaniolo, C.
Chien, S
Tsotras, V
Zaniolo, C
The VLDB Journal, Volume 11, Issue 4 (Dec. 2002)
Citations range: 
100 - 499
Chien2002Efficientschemesformanaging.pdf905.18 KB

Multiversion support for XML documents is needed in many critical applications, such as software configuration control, cooperative authoring, web information warehouses, and ”e-permanence” of web documents. In this paper, we introduce efficient and robust techniques for: (i) storing and retrieving; (ii) viewing and exchanging; and (iii) querying multiversion XML documents. We first discuss the limitations of traditional version control methods, such as RCS and SCCS, and then propose novel techniques that overcome their limitations. Initially, we focus on the problem of managing secondary storage efficiently, and introduce an edit-based versioning scheme that enhances RCS with an effective clustering policy based on the concept of page-usefulness. The new scheme drastically improves version retrieval at the expense of a small (linear) space overhead. However, the edit-based approach falls short of achieving objectives (ii) and (iii). Therefore, we introduce and investigate a second scheme, which is reference-based and preserves the structure of the original document. In the reference-based approach, a multiversion document can be represented as yet another XML document, which can be easily exchanged and viewed on the web; furthermore, simple queries are also expressed and supported well under this representation. To achieve objective (i), we extend the page-usefulness clustering technique to the reference-based scheme. After characterizing the asymptotic behavior of the new techniques proposed, the paper presents the results of an experimental study evaluating and comparing their performance.