MC2: MPEG-7 content modelling communities

Abstract

The use of multimedia content on the web has grown significantly in recent years. Websites such as Facebook, YouTube and Flickr cater for enormous amounts of multimedia content uploaded by users. This vast amount of multimedia content requires comprehensive content modelling otherwise retrieving relevant content will be challenging. Modelling multimedia content can be an extremely time consuming task that may seem impossible particularly when undertaken by individual users. However, the advent of Web 2.0 and associated communities, such as YouTube and Flickr, has shown that users appear to be more willing to collaborate in order to take on enormous tasks such as multimedia content modelling. Harnessing the power of communities to achieve comprehensive content modelling is the primary focus of this research. The aim of this thesis is to explore collaborative multimedia content modelling and in particular the effectiveness of existing multimedia content modelling tools, taking into account the key development challenges of existing collaborative content modelling research and the associated modelling tools. Four research objectives are pursued in order to achieve this; first, design a user experiment to study users’ tagging behaviour with existing multimedia tagging tools and identify any relationships between such user behaviour; second, design and develop a framework for MPEG-7 content modelling communities based on the results of the experiment; third, implement an online service as a proof of concept of the framework; fourth, validate the framework through the online service during a repeat of the initial user experiment. This research contributes first, a conceptual model of user behaviour visualised as a fuzzy cognitive map and, second, an MPEG-7 framework for multimedia content modelling communities (MC2) and its proof of concept as an online service. The fuzzy cognitive model embodies relationships between user tagging behaviour and context and provides an understanding of user priorities in the description of content features and the relationships that exist between them. The MC2 framework, developed based on the fuzzy cognitive model, is deep-rooted in user content modelling behaviour and content preferences. A proof of concept of the MC2 framework is implemented as an online service in which all metadata is modelled using MPEG-7. The online service is validated, first, empirically with the same group of users and through the same experiment that led to the development of the fuzzy cognitive model and, second, functionally against the folksonomy and MPEG-7 content modelling tools used in the initial experiment. The validation demonstrates that MC2 has the advantages without the shortcomings of existing multimedia tagging tools by harnessing the ease of use of folksonomy tools while producing comprehensive structured metadata.