| With the development and popularity of Web technology, the strong points of XML data as web information exchanging and storage media have been deeply rooted in people's mind. The native XML database specializing in XML data has drawn more and more attention with a promising future. The eXist database, which is regarded as the most popular open-source native XML database, has been running on almost all the platforms including embedded systems, and processing large quantities of real business data. Compared with other native databases, eXist database has higher efficiency in query, more integrated architecture as well as more affluent APIs. But when compared with mature rational databases, eXist database still has a gap in data query efficiency, data encryption management, and so on.This paper provides a detailed research on structural join algorithm in data query mechanism and encryption mechanism in data management of eXist database. In structural join algorithm, research has been done to address two problems: how to avoid the repeated traversal of node lists in structural join process, and how to decrease the unnecessary computing times to get matched results with current node encodings. In encryption mechanism, this paper mainly researches on how to provide an extensible mechanism for users to add new encryption conveniently without changing the database's current design. On the basis of the research above, this paper design and implement a fast locating structural join algorithm based on stack, which is called as FLStack, and a modularized encryption management mechanism featuring plug-in.Structural join operation is the core step of data query. This paper learns from another existing structural join algorithm called stack tree, which is more an algorithm rather than an application designed for a special database and has never been used in real databases. Based on the stack idea from stack tree algorithm and the current design of eXist database, this paper proposes a new structural join algorithm called Base_Stack to solve the problem of repeated traversal.However, in BaseStack algorithm, it still needs to compute the nodes' relations one by one before moving to the next potential ancestor or descendent node. As a result, this paper proposes another structural join algorithm featuring fast locating and stack, which is named as FL_Stack. By making use of the features from the current node encodings, FL_Stack defines the term of least differentiated prefix. By using the least differentiated prefix and the ancestor node as locating criteria, FL_Stack algorithm can skip the obviously non-matching ancestors and descendants, resulting in time saving in structural join operation. Meanwhile, FL_Stack algorithm is not only used in eXist database, but also can be applied to the applications with structural join operation.Then, this paper uses mathematics methods to prove the correctness from stack and fast locating points. Besides, it also implements lots of all directional experiments on synthetic data and real XML data collections. It is proved that FL_Stack algorithm has higher efficiency than the current structural join algorithm in eXist database.For the encryption mechanism in data security management, eXist database only provides MD5 encryption algorithm in its current version and doesn't support other encryption extensions. By defining a common interface for encryption algorithm and a respective encryption algorithm management module, this paper designs and implements an encryption management mechanism to add new encryption algorithm by modularization as a plug-in. This mechanism has improved the extensibility for encryption management in eXist database. At last, by adding a new encryption algorithm as an example, this paper describes the usage of this encryption management mechanism. Compared the encryption result with eXist database's current encryption algorithm, this encryption management mechanism is proved to function well in eXist database. |
PDF Full Text Request