Focus, salience, and priming in cyber-physical intelligence

Abstract

Information and language processing in a growing variety of computer applications can achieve the accuracy/precision and reliability that human users require if based on human-like understanding. While statistical and/or machine learning methods in text processing have been perfected in the last two decades (see, for instance, [1, 2] and references there), their successes have been limited in scope, primarily to the clustering tasks, and even there, the applications have excelled better in recall than in precision, leading to customer disappointments with commercial products.