We consider aspects pertinent to evaluating creativity to be input, output and the process by which the output is achieved. These issues

I summarise and attempt to clarify some concepts presented in and arising from Margaret Boden's (1990) descriptive hierarchy of creativity, by formalising the ideas she proposes.

Recently, many programs have been written to perform tasks which are usually regarded as requiring creativity in humans. We can

The purpose of this paper is twofold: to show the practical applications of theoretical evaluation measures designed to capture the degree of creativity of a program, and to use the results to evaluate an effort to develop an automatic Spanish poet. Existing efforts for the development of automatic poets are described, and the implications of their particular architectures to the evaluation issues discussed are considered.

Abstract The work of Boden on the nature of creativity has been extremely influential, particularly the hypothesis that the highest form of creativity results from transformation of a conceptual space. We consider how these ideas could be made more precise, and hence become amenable to empirical testing. This requires some reconsideration of foundational assumptions about computational creativity. We set down the abstract requirements for a conceptual space, review some possible types of formal model, and discuss how it might be possible experimentally to falsify (or corroborate) this hypothesis. We conclude that the central terms (conceptual space, transformation) are still too vaguely defined to support falsifiable claims, but that this is not an obstacle to writing creative computer programs.

Abstract. We investigate the automatic construction of visual scenes via a hybrid evolutionary/hill-climbing approach using a correlationbased fitness function. This forms part of The Painting Fool system, an automated artist which is able to render the scenes using simulated art materials. We further describe a novel method for inventing fitness functions using the HR descriptive machine learning system, and we combine this with The Painting Fool to generate and artistically render novel scenes. We demonstrate the potential of this approach with applications to cityscape and flower arrangement scene generation. 1

This paper examines two seemingly unrelated qualitative spatial reasoning domains; geometric proportional analogies and topographic (landcover) maps. We present a Structure Matching algorithm that combines Gentner’s structure mapping theory with an attribute matching process. We use structure matching to solve geometric analogy problems that are centered on manipulating attribute information, such as colors and patterns. Structure matching is also used to creatively interpret topographic (land-cover) maps, serving to add a wealth of semantic knowledge and providing a far richer interpretation of the raw data. We return to the geometric proportional analogy problems and identify alternate attribute matching processes that are required to solve different categories of geometric proportional analogy problems. Finally, we assess some implications for computationally creative and inventive models.

The shape taken by linguistic creativity at the different levels of decision involved in sentence production (phonetics, rhythm, lexical choice, semantics, syntax and narrative content) is explored in relation to existing computational models of creativity. A general outline of the possibilities is given for each level, and two specific levels - word invention at the lexical level, illustrated by the Jabberwocky poem by Lewis Carroll; and poetic metaphor at the semantic level, illustrated by examples from verses by Garcia Lorca - are studied in further detail. The applicability of the existing computational models is discussed in connection to the kind of creativity apparent in the examples.

creativeness / a pleasing field / of bloom Word associations are an important element of linguistic creativity. Traditional lexical knowledge bases such as WordNet formalize a limited set of systematic relations among words, such as synonymy, polysemy and hypernymy. Such relations maintain their systematicity when composed into lexical chains. We claim that such relations cannot explain the type of lexical associations common in poetic text. We explore in this paper the usage of Word Association Norms (WANs) as an alternative lexical knowledge source to analyze linguistic computational creativity. We specifically investigate the Haiku poetic genre, which is characterized by heavy reliance on lexical associations. We first compare the density of WAN-based word associations in a corpus of English Haiku poems to that of WordNet-based associations as well as in other non-poetic genres. These experiments confirm our hypothesis that the non-systematic lexical associations captured in WANs play an important role in poetic text. We then present Gaiku, a system to automatically generate Haikus from a seed word and using WAN-associations. Human evaluation indicate that generated Haikus are of lesser quality than human Haikus, but a high proportion of generated Haikus can confuse human readers, and a few of them trigger intriguing reactions.

We begin with a number of basic facts about creativity and a brief history of the idea. These provide criteria that any definition of the term should meet and help guide us to a new definition of creativity. This definition is independent of cultural appropriateness or the perceived value of creative objects, ideas which have encumbered previous investigations. We briefly defend our definition against some plausible objections and then explore the ways in which this new definition differs from alternative views, by improving upon them.