Institutional Repository

Conceptual blending techniques for data visualisation

Show simple item record

dc.contributor.advisor Van der Poel, Etienne
dc.contributor.author Featherstone, Coral
dc.date.accessioned 2019-10-15T09:03:21Z
dc.date.available 2019-10-15T09:03:21Z
dc.date.issued 2018-09
dc.identifier.uri http://hdl.handle.net/10500/25851
dc.description.abstract Computational creativity is an active area of research within the artificial intelligence domain that investigates what aspects of computing can be considered as an analogue to the human creative process. Computers can be programmed to emulate the type of things that the human mind can. Artificial creativity is worthy of study for two reasons. Firstly, it can help in understanding human creativity and secondly it can help with the design of computer programs that appear to be creative. Although the implementation of creativity in computer algorithms is an active field, much of the research fails to specify which of the known theories of creativity it is aligning with. The combination of computational creativity with computer generated visualisations has the potential to produce visualisations that are context sensitive with respect to the data and could solve some of the current automation problems that computers experience. In addition theories of creativity could theoretically compute unusual data combinations, or introducing graphical elements that draw attention to the patterns in the data. More could be learned about the creativity involved as humans go about the task of generating a visualisation. The purpose of this dissertation was to develop a computer program that can automate the generation of a visualisation, for a suitably chosen visualisation type over a small domain of knowledge, using a subset of the computational creativity criteria, in order to try and explore the effects of the introduction of conceptual blending techniques. The problem is that existing computer programs that generate visualisations are lacking the creativity, intuition, background information, and visual perception that enable a human to decide what aspects of the visualisation will expose patterns that are useful to the consumer of the visualisation. The main research question that guided this dissertation was, “How can criteria derived from theories of creativity be used in the generation of visualisations?”. In order to answer this question an analysis was done to determine which creativity theories and artificial intelligence techniques could potentially be used to implement the theories in the context of those relevant to computer generated visualisations. Measurable attributes and criteria that were sufficient for an algorithm that claims to model creativity were explored. The parts of the visualisation pipeline were identified and the aspects of visualisation generation that humans are better at than computers was explored. Themes that emerged in both the computational creativity and the visualisation literature were highlighted. Finally a prototype was built that started to investigate the use of computational creativity methods in the ‘variable choice’, and ‘aesthetics’ stages of the data visualisation pipeline. en
dc.format.extent 1 online resource (146 leaves) : illustrations (some color)
dc.language.iso en en
dc.subject.ddc 006.37
dc.subject.lcsh Information visualization -- Data processing
dc.subject.lcsh Data mining -- Graphic methods
dc.subject.lcsh Visual programming (Computer science)
dc.subject.lcsh Computer vision
dc.title Conceptual blending techniques for data visualisation en
dc.type Dissertation en
dc.description.department School of Computing en
dc.description.degree M. Sc. (Computing)


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search UnisaIR


Browse

My Account

Statistics