Wang, Feng2024-08-152024-08-1519961996https://hdl.handle.net/1828/20039In this thesis, we present a multidimensional spreadsheet-style browser. This browser allows users to visualize and analyze multidimensional data sets, in particular multidi­mensional output data produced by Granular LUCID (GLU). As GLU has simplified the complex task of generating multidimensional data, the need for an easy and efficient means of viewing these data becomes more evident. GLU is a hybrid of the LUCID and C languages. LUCID is the declarative nucleus of GLU. Because LUCID is a functional, dataflow language for processing multidimen­sional data, the traditional imperative approach to I/O is not appropriate. Instead, a multi­dimensional browser can be used to display and analyze these multidimensional output data sets generated by GLU. The approach we develop for browsing the output of multidimensional GLU pro­grams is a three step process. First, all output data of the multidimensional GLU program are collected in a multidimensional warehouse file. Second, all data in this warehouse are read into memory. Finally, users browse these data from a variety of viewpoints. In addition to a tabular view of the multidimensional data set, the browser provides several analysis functions. Users select data for further investigation by mapping two chosen dimensions to the horizontal and vertical axis of the display and fixing all other dimensions. Given these parameters, the browser can generate a spreadsheet view accord­ing to the user's need. For dimensions with a large coordinate space, users can zoom in a particular portion of the display for combination and comparison. Furthermore, common mathematical functions such as SUM, AVE, MIN, MAX can be performed. These types of features are often supported by on-line analytical processing (OLAP) products. Like any OLAP product, this browser provides three basic functions - slicing and dicing, comparison and combination of rows and columns, and analytical calculations. In order to effectively achieve these functions for browsing multidimensional data our design involves careful consideration of a storage structure, storage order and data access address scheme. Attention to these key elements yields a user friendly browser for use with any multidimensional data warehouse of similar structure.80 pagesAvailable to the World Wide WebThesis