Use this URL to cite or link to this record in EThOS:
Title: Distributed decision support systems
Author: Alkarouri, Muhammad Abdulmuneim
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2010
Availability of Full Text:
Access from EThOS:
Decision support systems are a class of computer based systems that assist in some or all levels of decision making within an organisation. Recently, the growth of data captured that is useful or even critical to the successful running or conclusion of projects in science and industry has been remarkable. Thus, the development of decision support systems that are scalable in terms of the size of data processed. the number of stakeholders, and their geographical span has become of the essence. This thesis identifies the issues in developing distributed decision support systems. Building on that. an architectural style for the development of scalable and extensible software systems is introduced. Subsequently, a framework for the design of distributed decision support systems is developed. This new architectural style is the Resource Oriented Services Architecture (ROSA). It builds on Representational State Transfer (REST), an architectural style that describes the venerable design of the world wide web. An architectural design based on REST revolves around resources, representations, and hyperlinks. \Vhat it lacks is a standardised way to represent computations as resources in a scalable and extensible manner. For systems that cannot be adequately described as a web of documents, this is a shortcoming. ROSA overcomes this by defining a means of representing executable resources in a manner that is consistent with the statelessness and cacheability constraints of REST. The resulting architecture enables the scalability of the system. Additionally, desirable features such as dynamic discovery of resources and extensibility and loose coupling are attained. To illustrate this framework, two new learning algorithms are introduced and implemented as services. The first is a data structure suitable for proximity queries over large datasets of low intrinsic dimension. The other uses a random projection to carry out novelty detection over high dimensional datasets.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available