Incremental Structured Index Update Mechanism and Path Expression Query for XML Documents

• Main Authors: Chien-Che Hung, 洪健哲
• Other Authors: Hao-Ren Ke
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/31955121111529273317

碩士 === 國立交通大學 === 資訊科學系 === 90 === XML documents contain abundant structural information that can be employed to better understand the XML documents. Structured queries are queries that search the structural information of XML documents. In order to support structured queries, k-ary trees are used in this thesis as an index structure to store structural information. Based on this index structure, this thesis proposes an incremental structured index update mechanism and implements a few functions of a SQL-like path expression query language, called Lorel. K-ary trees are a well-known index structure can be used to store structural information of XML documents, where k is a predefined maximum number of branches that a tree node can have. When an XML document is modified, its corresponding k-ary tree has to be updated as well for reflecting the modification. However, when a modification makes the number of branches of a tree node exceeds k, the k-ary tree is no longer workable and the whole k-ary tree has to be reconstructed. Reconstructing the whole k-ary tree is time-consuming when the structure information changes frequently. This thesis proposes an incremental structured index update mechanism called the Split Method to avoid reconstructing the k-ary tree when the number of branches of a node exceeds k. Through splitting nodes and defining an equivalent relationship, this mechanism can update k-ary trees efficiently. Several evaluations were conducted to assess the Split Method and the k-ary tree reconstruction. The evaluation shows that the Split Method only wastes a little more space than tree reconstruction, but it is more efficient than tree reconstruction in updating k-ary trees. In addition, based on the k-ary tree structured index, this thesis implements a few functions of a SQL-like path expression query language, called Lorel. Path expression queries facilitate user to issue structural queries when the document structure is irregular or unknown in advance. The results of queries are formatted as XML documents.