Oxone: A Scalable Solution for Detecting Superior Quality Deltas on Ordered Large XML Documents

Abstract

Recently, a number of relational-based approaches for detecting the changes to XML data have been proposed to address the scalability problem of main memory-based approaches (e.g., X-Diff, XyDiff). These approaches store the XML documents in the relational database and issue SQL queries (whenever appropriate) to detect the changes. In this paper, we propose a relational-based ordered XML change detection technique (called Oxone) that uses a schema-conscious approach as the underlying storage strategy for XML data. Previous efforts have focused on detecting changes to ordered XML in an schema-oblivious storage environment. Although the schema-oblivious approach produces better result quality compared to XyDiff (a main memory-based ordered XML change detection approach), its performance degrade with increase in data size and is slower than XyDiff for smaller data set. We propose a technique to overcome these limitations. Our experimental results show that Oxone is up to 22 times faster and more scalable than the relational-based schema-oblivious approach. The performances of Oxone and XyDiff (C version) are comparable. However, more importantly, our approach is more scalable compared to XyDiff for larger datasets and has much superior the result quality of deltas than XyDiff.