A Big Data and Darwinian Approach of Scientific Creativity

摘要

At present time, we can find more than fifty frameworks for creativity management. Some call themselves theories; some of them prefer to be known as methods. Creativity remains today a major pre-requisite for innovation, but remains mostly unknown, as well as the entrepreneurship process itself. Some recent studies even statistically experimented a link between creativity and some mental disorders (Kyaga,2012). Using the entrepreneurship process and the Saras Sarasvathy's effectuation approach as a reference about human behaviour under uncertainty, we'll analyse the connection with another Indian approach of practical creativity called 'Jugaad'. Most of the methods available, mainly the ones closely linked to psychology, are not skill oriented. However, the idea generation process is useless if disconnected with the ability to perform relevant and cost efficient proof-of-concepts, including experimental design approaches, experimental accuracy and precision assessment, hidden parameters detection... In this paper, we will propose a state of the art regarding what we know about creativity conditions and what is the typology of the creativity frameworks (psychology, linguistic, management) since 1960. The meaning of innovation evolved in western countries from a concept basically linked to invention and practical creativity, as Jugaad proposes, to a concept linked to participative innovation, design thinking, user experience approach. The first one highlights individual action, the second one implies a collaborative process including an extreme variety of stakeholders. These two approaches have both advantages and drawbacks. We will study how it is possible to aggregate previous experiences using big data technologies, semantic analysis and a Darwinian approach of human creativity. Based on an automated analysis of five millions scientific reports, we use a semantic synset data mining approach and a graph visualization to build a graphical representation of time-resolved concepts and knowledge. We'll present several applications and implications that have to be taken into account for an improved R&D management towards applied sciences. We especially worked on graphical representation of the technical knowledge, trying to discover a balanced way between accuracy and inspiration, between individual inventions and collaborative massive interaction.