Combinational creativity and computational creativity

Abstract

Creativity is a significant element of design. However, it can be challenging to produce creative ideas that will be of value to society and worthy of design effort. This thesis explores combinational creativity, which involves unfamiliar combinations of familiar ideas, and implements combinational creativity in computational tools to support design. Two computational tools, the Combinator and the Retriever, have been developed to support designers in creative idea generations during the early stages of design. The Combinator has been developed by imitating aspects of human cognition in achieving combinational creativity. This tool produces combinational prompts in text and image forms. The Retriever has been developed based on ontology by embracing aspects of cognitions of analogical reasoning. This tool constructs ontologies with sufficient richness and coverage to support reasoning over real-world datasets, and produces text-form outputs with correlated image mood boards. Case studies indicate that the Combinator and the Retriever are useful and effective in terms of supporting creative ideation. The case studies show that both of the tools can increase the fluency of idea generation and improve the flexibility, usefulness and originality of the ideas produced, for the datasets studied. In addition to computational creativity tools, this thesis has proposed three driven approaches to produce combinational creative ideas, which are problem-, similarity-, and inspiration-driven. A study indicates that these three approaches are used commonly in practical designs, of which the problem-driven approach is the dominant approach. In addition, the three approaches could be used individually as well as in groups. The conceptual distances between ideas in combinational creativity are also explored in this thesis. A study reveals that far-related ideas, which are used more commonly, could lead to more creative outcomes compared with closely-related ideas. The findings and outcomes of this thesis lead to a further understanding of creativity in the design context. The novel approaches used for developing the Combinator and the Retriever, as well as the outcomes of the theoretical studies could be adapted to develop new computational design support tools.