The preference update framework for Web and e-commerce applications.

详细信息   

• 作者：Li ; Peng.
• 学历：Doctor
• 年：2005
• 导师：Yen, I-Ling
• 毕业院校：The University of Texas
• 专业：Computer Science.
• ISBN：0496971204
• CBH：3163250
• Country：USA
• 语种：English
• FileSize：1094351
• Pages：162

文摘

Transactions in e-commerce and Web-based applications frequently involve databases that contain consumable resources, such as limited number of items to be sold. When a customer issues such a transaction based on out-of-date information, the transaction may fail because the desired data resources are no longer available. Subsequently, the customer has to issue another transaction for alternative selections. This process may occur repeatedly, resulting in low customer satisfaction and high communication cost. Since Web-based and e-commerce application systems usually have a high usage and access volume, their data states change rapidly. This exacerbates the request failure problem.;In this dissertation, we develop a novel preference update framework to address the above problem. The core of this framework is the preference update model. This model extends the standard SQL update statement in three aspects: (a) Multiple preferences can be defined within one update statement, where each preference is defined by a logical formula to express the search conditions; (b) Multiple preferences within one statement can be defined over different relations; (c) The cardinality of result set can be specified in a highly flexible way. With this model, a customer can specify her request in a batch way instead of iteratively submitting alternative statements in the case of failures. The system processes a preference update request by updating the most preferred data items available in the databases according to the preferences.;Based on the preference update model, we develop the supporting language, the realization system architectures, the optimization schemes, and the group evaluation strategy. The realization system architectures include tightly and loosely coupled styles. For each architecture, we develop several alternative approaches for query execution, show how to evaluate the costs of these approaches and how to optimize the query processing. The group evaluation strategy makes optimal data allocation decisions for multiple concurrent preference update requests to further reduce the transaction failure rate and improve customer satisfaction.