Extracts cognitive artifacts from text through combining human and machine learning in an iterative fashion

Abstract

The world network of information is complex and not always organized in a structure useful for human understanding. This paper investigates the need and methods for creating an artificial system that categorizing information similar to the way humans categorize. The system will use Bayesian modeling to model text sentiment. The categorization of text sentiment will be done both by machines and by humans. The hypothesis is that the resultant system will not differ significantly from the accuracy of a control group of human categorizers. This represents a non-standard approach to learning that involves the human and the machine in an iterative learning process. © 2011 Springer-Verlag.