Physical Sciences and Mathematics Commons^™

An Efficient Implementation Of Logical Design For Relational Databases, Shakeel Ishaque

Masters Theses

This work presents efficient methodology for the design of relational databases and an implementation of a design tool. A set of algorithms and supporting theory are discussed. Improvements are made on existing decomposition approaches.

The Dependency Preserving Normal Fom (DPNF), which is stronger than 3NF, is presented. It guarantees a decomposition with lossless join property and at the same time preserves all of the functional dependencies.

An algorithm to obtain DPNF decomposition is presented. The algorithm computes DPNF decomposition in polynomial time. It converts supplied functional dependencies to annular cover and uses reduced annular cover to compute closures and decomposition. …

A Generalized Data Distribution Model For Distributed Relational Databases, Patrick D. Yurk

Masters Theses

This thesis presents a generalized model for the allocation of data in a distributed relational database. The model is applicable to a network of computers where there are different retrieval and update costs at the various sites, and different communication costs between sites.

The method used in this model is based upon clustering of similar computers within the network. Clusters are derived from grouping together sites with similar update, retrieval and communication costs. Data is then allocated to a cluster based on actual communication and processing costs.

The data distribution method between clusters uses a technique where the benefit of …

The Implementation Of A Generalized Database Directory, Claudia Helms-Vaccaro

Masters Theses

This paper presents the use of an index made up of a single B+ tree connected at its leaf nodes to a lower set of B+ trees--a multitree. It may be utilized to index dense or non-dense attributes of a file. An extra benefit of this structure is the ability to store a secondary attribute value for use when searching the index, often decreasing accesses and time required to complete the search.

Improved efficiency over current indices, such as conventional B+ trees or inverted files, is evaluated.

An Analysis Of High Dimensional Data, Ravishankar Swamy

Masters Theses

Projection of a high dimensional data set into a two dimensional subspace can lead to very interesting projections. Note that P dimensional data can be projected onto P orthogonal two dimensional planes, which completely determines the original data. Two dimensional orthogonal projections can be analyzed for patterns in several ways.

In this work we will be trying to explore an unknown P dimensional data set by comparing its test projections with a set of standard patterns. A set of P dimensional data is transformed into K.K grid counts and the whole process of recognition is done using these K.K grid …