| Summary: | 博士 === 國立成功大學 === 資訊工程學系碩博士班 === 98 === Supporting user preferences has recently been a growing interest in extending the capability of database systems. In our work, we propose several efficient structures and algorithms for evaluating such user preference queries in centralized and distributed environments. Since some of the user preference queries are processed in distributed sensor networks which are usually with limited power sources and low computation capability, we first dynamically adjust its sensing radius in our work so that the global coverage of the whole detecting area remains unchanged, energy consumption is minimized, and the lifetime of the sensor network is extended. Specifically, we study the evaluation of three specific types of user preference queries. First, a similarity query allows the users to retrieve the similar data, even if there is no value which exactly matches the given value of the query in sensor networks. Second, a skyline query allows the users to specify whether they favor low, high or different values of the attributes in sensor networks. Third, a top-k combinatorial skyline query allows the users to evaluate the top-k interesting skyline combinations from various kind of combinations of the give tuples. We propose several approaches to efficiently evaluate these queries. All approaches are progressive and provide a fast initial response time. All the schemes are further analyzed empirically through extensive experimental studies and the results indicate that they are effective in supporting user preferences in both centralized and distributed database environments.
|
|---|
