Knowled0e-l~cl 5V$TEM$"-ELSEVIER
Knowledge-BasedSystems9 (1996) 349-350
Preface
Creativity and Cognition 1996 Creativity and Cognition 1996 marked a consolidation of the first symposium in 1993 from which a selection of papers were reported in Knowledge-Based Systems [1]. This special issue of Knowledge-Based Systems consists of a selection of revised papers that were presented in an earlier form at the 1996 Symposium. These papers raise a number of new questions about the nature of human creativity and the role of computer systems in modelling and supporting the creative process. In particular, the part that knowledge plays in creativity is proving to be a key issue. Knowledge applied in creativity, its character, form and how we might define its role in computer support, are research issues that are being tackled in a number of ways. The need to identify different categories and levels of knowledge about the domain of concern and how it might be represented in system design are key challenges. Another related concern is the reflexive relationship between existing knowledge and emerging knowledge structures and the implications for the constructions of computer support systems. Given a knowledge system that evolves with its user's knowledge or" incorporates responsive elements, the issue being addressed is the dynamic nature of both the human and the system knowledge. Designing computer support environments that support different user knowledge strategies is a promising research area. When such technologies become mature and are used extensively in creative work, they will present interesting scenarios about the role of computers in creative work. For support to be truly helpful there must be sufficient support knowledge in the system and the means to activate it automatically or control it directly. The allocation between autonomous functions and user-directed functions is a matter that engages the designers of interactive computer systems. There are claims that the computer system changes and, indeed, enhances human creativity. This issue is one that will become more significant as and when support systems become smarter and really usable and appropriate for creative work done in anger. The design of potentially creativity supporting computer systems is now firmly on the research agenda. The challenge 0950-7051/96/$15.00 ,!) 1996 Elsevier Science B.V. All rights reserved Pll S0950-7051 (96)01056-8
will be to create support environments that go well beyond current concerns for better interaction techniques and the emulation of human cognitive processes. In attempting to conduct research on these themes the international body of researchers associated with Creativity and Cognition are pushing our understanding forward both in terms of understanding creativity better and in terms of knowing how to support it. Ambach and Repenning describe a fine example of emerging technologies that, as they move towards fully fledged systems, will pose interesting questions about the role of computers in creative work. Making computer software environments for users to explore art in a way that demands more active participation on the part of the computer system is a key motivation for this work. The authors put forward the view that existing tools are too passive, and advocate bringing tools that have autonomous capability into human exploratory activities. Users become "Directors" rather than "Puppeteers". The allocation of some aspects of human exploration, they claim, would be enhanced by creating tools with more active roles. A software environment, Agentsheets, which enables users to adopt different degrees of control, is described. The crucial feature of the system is that of delegation, where users direct the process rather than control it, for as they argue, having total control usually means a high degree of effort on the part of the user. The creation of Agentsheets has enabled the authors to observe the strategies adopted by users and also the changes that take place from early exploratory behaviour leading typically to random output towards selfimposed development of constraints and, hence, the production of a more controlled output. If this is a consistent effect of the system on users, there would seem to be a case for claiming that the system encourages more creative behaviour for, as Boden has sought to demonstrate, introducing and changing constraints may be a central facet of creativity [2]. The primary concern of the Bruderlin paper is to develop computer tools for supporting the human animator. The difference from other animation software programs is that he is developing tools that do not
350
L. Candy~Knowledge-Based L~vstem,~" 9 /1996 ) 349 350
depend upon low level functions for creating and composing human figures. If the animation process is to be supported fully, the tools must incorporate knowledge about movement and motion in order to facilitate the motion specification task. The concept of human figure animation as primarily about motion control is an important assumption in this work. An interactive system should support the animator in developing the motion ideas in a flexible way without removing the necessary control and creativity. Usually there is a trade-off between what is embedded in the system about motion and how much the animator is allowed to specify and control. Some of the questions he poses in this area imply a need for several levels of knowledge about the domain and about the processes of achieving the animation. Most animation systems still rely on lowlevel keyframing which provides detailed control but can be tedious to use. What is really required is a multilayered approach to motion control which allows the animator to specify and modify at any level in the motion hierarchy. This motion hierarchy allows for creation and modification at different levels of abstractions or granularities of animation. Every level of the motion control hierarchy needs tools that constitute high level building blocks for the creative process. The paper by Sugimoto, Hori and Ohsuga is concerned with support for creative concept formation. A system for assisting researchers' creativity by visualizing different viewpoints on a certain research topic is described. The system has a text database composed of journal and conference papers, and can elicit the viewpoints of the papers automatically and visualize the semantic relations between them. The aim is to reveal the structure behind the users' mental worlds and to enhance their understanding of the subject area. User studies were conducted from which it was concluded that finding relations between things not previously found took place. In other words, assistance was provided to make tacit knowledge explicit. Support for creativity lies in enabling people to see things not previously noticed or to remind them of what they have forgotten. The claim made by the authors is that they have developed a tool that not only aids communication and information retrieval but has the potential for amplifying human intelligence. The papers by Nishimoto, Sumi and Mase and Schmid are examples of work concerned with modelling human creativity and making computers behave creatively. Nishimoto, Sumi and Mase's start point is the value of conflict between different concepts to creativity and the role of brainstorming in stimulating divergent thinking. The problem identified is that normally people brainstorm within a common frame of reference and there are, therefore, limited opportunities for drawing upon knowledge outside that frame. The creation of
systems that act as agent outsiders which bring to bear domain knowledge from several different sources offers potential for creativity. To achieve this, the computer system needs to be able to derive its knowledge from other sources and to synthesize the information it acquires. The really smart system must be able to filter that which might appear to be irrelevant to human experts. From this concept, an outsider model has been derived, a prototype system developed and experiments conducted to evaluate the basic characteristics of the system. Schmid proposes a model of information processing in human creativity called the IPC model. The model is based on psychological studies and comparable models and tries to capture different types of knowledge that operate in creativity. For example, he distinguishes between active and inactive knowledge, internal and external knowledge and meta-knowledge. Based upon the role of knowledge in creativity, he derives a set of assumptions about knowledge representation, for example a need to have both declarative and procedural knowledge and to be able to translate one from the other. He uses the model as a baseline for assessing existing artificial intelligence approaches and discusses ways of providing AI systems with a higher degree of creativity. Interest in creativity, either from a research perspective or from the practitioner's point of view, does not bring with it a set of agreed goals and methods which mark the conventional well-established academic discipline. The papers in this special issue approach the subject from diverse viewpoints and with different goals. Whilst there is continued activity in the field of what might be termed the emulation of creative processes by computational means, new approaches to overcoming the limits of the old AI systems imply revisiting the theoretical foundations of the earlier work. In addition, more attention is being given to the process aspects of human tasks, skills and behaviour rather than the products of creative work as indications of creative behaviour in the machine. Focusing upon human creativity brings with it a concern for the relationship between its various parts, and thus the dynamics of learning and the role of knowledge are brought into the frame of reference.
Linda Candy LUTCHI Research Centre, Loughborough, UK
References [1] Creativity and Cognition, Special issue of Knowledge-Based Systems,Volume7, No 4 (December 1994). [2] M.A. Boden, The Creative Mind: Myths and Mechanisms, Weidenfeldand Nicolson, London, 1990.