From Information to Intelligence: The Role of Relative Significance in Decision Making and Inference

The growth of the amount of information accessible to people is phenomenal in recent years. In one estimate, the amount of digital information created, captured, and replicated in 2006 was 161 exabytes, approximately three million times the information in all the books ever written [1]. The issue of finding or discovering intelligence from a sea of information thus becomes ever more important and challenging. Although computational algorithms have been developed to aid the user in search of information that may be deemed useful, automatic discovery of intelligence (ADI) beyond simple information matching is still beyond the reach of conventional search methods. In this paper, we address one important component, namely relative significance which is essential in forming the paradigm of intelligence beyond a simple recognition of pattern of information. We further formulate the concept of relative significance in the framework of Bayes decision theory so as to enable computational optimization and algorithmic development. We believe processing of true computational intelligence must include capabilities in dealing with nonuniform significance and cost issues.