Limiting context by using the web to minimize conceptual jump size

Abstract

In this paper we introduce our ideas on how experiences from real situations could be processed to decrease what Solomonoff called "Conceptual Jump Size". We introduce applications based on commonsense knowledge showing that vast corpora are able to automatically confirm the validity of the output, and also replace a "trainer", which could lead to decreasing human influence and speeding up the process of finding solutions not provided by such a "trainer" or by real world descriptions. Following this idea, we also suggest a shift toward combining natural languages with programming languages to smoothen transitions between layers of Solomonoff's "Concept net" leading from primitive concepts to a problem solution. © 2013 Springer-Verlag Berlin Heidelberg.