Context-based multimedia semantics modelling and representation

Eze, Emmanuel Uchechukwu

Abstract

The evolution of the World Wide Web, increase in processing power, and more network bandwidth have contributed to the proliferation of digital multimedia data. Since multimedia data has become a critical resource in many organisations, there is an increasing need to gain efficient access to data, in order to share, extract knowledge, and ultimately use the knowledge to inform business decisions. Existing methods for multimedia semantic understanding are limited to the computable low-level features; which raises the question of how to identify and represent the high-level semantic knowledge in multimedia resources.

In order to bridge the semantic gap between multimedia low-level features and high-level human perception, this thesis seeks to identify the possible contextual dimensions in multimedia resources to help in semantic understanding and organisation. This thesis investigates the use of contextual knowledge to organise and represent the semantics of multimedia data aimed at efficient and effective multimedia content-based semantic retrieval.

A mixed methods research approach incorporating both Design Science Research and Formal Methods for investigation and evaluation was adopted. A critical review of current approaches for multimedia semantic retrieval was undertaken and various shortcomings identified. The objectives for a solution were defined which led to the design, development, and formalisation of a context-based model for multimedia semantic understanding and organisation. The model relies on the identification of different contextual dimensions in multimedia resources to aggregate meaning and facilitate semantic representation, knowledge sharing and reuse. A prototype system for multimedia annotation, CONMAN was built to demonstrate aspects of the model and validate the research hypothesis, H₁.

Towards providing richer and clearer semantic representation of multimedia content, the original contributions of this thesis to Information Science include: (a) a novel framework and formalised model for organising and representing the semantics of heterogeneous visual data; and (b) a novel S-Space model that is aimed at visual information semantic organisation and discovery, and forms the foundations for automatic video semantic understanding.

Citation

Eze, E. U. (2013). Context-based multimedia semantics modelling and representation. (Thesis). University of Hull. Retrieved from https://hull-repository.worktribe.com/output/4214911

• Thesis Type: Thesis
• Deposit Date: Jan 20, 2014
• Publicly Available Date: Feb 23, 2023
• Keywords: Computer science
• Public URL: https://hull-repository.worktribe.com/output/4214911
• Additional Information: Department of Computer Science, The University of Hull
• Award Date: May 1, 2013