Oral Candidacy - Holly T. Ferguson
|Start:||4/15/2016 at 1:00PM|
|End:||4/15/2016 at 3:30PM|
|Location:||258 Fitzpatrick Hall|
Faculty and students are welcome to attend the presentation portion of the defense.
Holly T. Ferguson
April 15, 2016
258 Fitzpatrick Hall
Adviser: Dr. Jaroslaw Nabrzyski
Dr. Michelle Cheatham Dr. Tracy Kijewski-Correa Dr. Greg Madey
“AN ONTOLOGY-BASED APPROACH FOR DISTRIBUTED DYNAMIC DECISION SUPPORT INTEGRATION VIA ENVIRONMENTAL AND RISK ANALYSIS METHODOLOGIES”
As the quantities of data expand and complexity of data increases, methods for structuring and handling it efficiently are more vital than they have ever been. The ongoing development and integration of modern applications with the semantic web of data means that to reach its full potential, our data sets need to be as interoperable, accessible, and reliable as possible. However, as professionals continue to build applications which need an increasing amount of interdisciplinary data, there is also a need to build data patterns that can interconnect these concepts and relate them across a variety of sources. Additionally, effective integration methods are needed to align intelligent data sets into tools per availability, preferably in at least a semi-automated manner so as to only enhance the capabilities of associated decision support methods.
Thus far, our research has produced several of the required data patterns, explored semantic rules engines, and given literature reviews of the technological landscape as it exists today. This work is presented and explained relative to its importance in the larger scope of data and system interoperability. These individual projects provide a substantial foundation for the proposed work that is to take place in the following months to further assist the transitions between open linked-data and the decision support frameworks that need to efficiently make use of and readily update changes within the structured data over distributed data access points.
This proposed research seeks to overcome certain limitations of the typically proprietary nature of architectural applications which use numerous geometry and spatial schemas along with localized data sets and connect them together using Linked Data principles with Semantic Web endpoints to establish solutions within this domain that will be reusable for many domains and computer science research efforts. The aim is to develop the foundation framework of a cloud-based cyber-infrastructure that will be used to host our use-case data and simulation models relevant to the specific domains of architecture, resilience tools, and risk analysis techniques. This particular cyber-infrastructure is a type of Linked Data Platform and is crafted to be appropriate for distributed processing and robust decision support techniques that are becoming essential for handling Big Data (from our use case domains or any other) and processing it at several locations, often simultaneously.
The first stages of developing this infrastructure include plans for layering Ontology Design Patterns and proprietary schema translation techniques to eventually better-enable multi-criteria analysis over a diverse set of data types from diverse sources. These efforts have already begun in the form of a Linked Data schema adapter Pattern (called an Upper Spatial Ontology) as well as HYDRA-based REST interfaces for platform communications across distributed networks. This allows the next steps of adding these components to Docker containers and adding additional processing components as they become available, such as triple stores, reasoners, and a dynamic Knowledge Base. We intend to be able to establish automated hierarchies of data, computational actions for simulations, and user preferences, as well as to automate a knowledge base as information becomes available. These pieces will help to expand the functionality the cyber-infrastructure and enable the following set of components which would include creating a fully automated Knowledge Discovery and Decision Engine.