Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.769736
Title: Tracking in information space
Author: Greenhough, Helen
ISNI:       0000 0004 7659 1281
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
A long-term ambition of researchers is to develop readily available expert systems to support decision makers. There is great interest in combining web-based information and Machine Learning techniques, which are enjoying a renaissance attributed to the ready availability of computing resources, to answer real-world strategic questions. Given the contribution of industry to overall national economic health, one such question is what industry mix is necessary to support national strategy? The proposed conceptual graph model and accompanying framework, applied the strengths of machines and computational techniques to inform national industrial policy in a data-driven paradigm. This thesis makes three contributions: the development of a novel conceptual graph model and associated framework, both based on a multi-disciplinary literature review, and their subsequent demonstration using both simulated binary and real-world statistical data. The graph model, drawing from experience in the government defence sector, linked national intent to industry via intermediate nodes such as capability. The wider framework combined supervised Bayesian Belief Network Machine Learning with web-scraping via the graph model to infer national intent. Proof of concept was achieved through the use of both simulated and real-world data, confirming the ability of the chosen inference approach to learn the required relations. Ultimately the framework was able to identify industry variables which best explain observed capability for two real-world case studies although it failed to draw inferences across industries using real data. Further avenues of investigation were addressed, including the use of unstructured text as a data source and alternative Machine Learning methods to account for time-series data. The potential of the web as a data source is truly staggering. This thesis goes some way to bridging the gap between this vast store of information and its incorporation in an evidence-driven pipeline in support of strategic decision making.
Supervisor: Hankin, Chris ; Huth, Michael Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.769736  DOI:
Share: