| Summary: | 碩士 === 國立臺北科技大學 === 資訊工程系所 === 99 === In the study of data management, we typically assume that the information is absolutely accurate. However, in some environments, the information obtained is not always accurate. For example, the data we derive in a wireless sensor network. The data item we derive can be an estimate or statistics (e.g. Mean, variance, etc.). We refer to such data as the uncertain data. The uncertain data can be distributed according to some probability distribution. For instance, the Gauss distribution. In this thesis, we consider how to manage the uncertain data effectively. To search on the uncertain data, one can do the linear search to find the best possible answers but it is time consuming in a large uncertain data set. We propose index structures to index the uncertain data and the proposed index structures can support many kinds of queries, including point query, range query, and Top-k query. The query processing for each of the mentioned types of queries and the related operations, like insertion, deletion, and update, will be discussed and analysis. We furthermore provide the simulation experiments for the proposed schemes and have a comparison with the previous MV-tree index structure. The experimental results show that our index structures are not only efficient in terms of the number of I/O’s and accuracy, but also easier to maintain.
|
|---|
