Introduction

Minsky and Papert (1969) in their progressive and well-developed writing discussed the need to construct arti€cial intelligence (AI) systems from diverse components: a requisite blend of symbolic and connectionist approaches. In the symbolic approach, operations are performed on symbols, where the physical counterparts of the symbols, and their structural properties, dictate a given system’s behavior (Smolensky, 1987; Spector, 2006). It is argued that traditional symbolic AI systems are rigid and specialized, although there has been contemporary development of symbolic “learning” systems employing fuzzy, approximate, or heuristic components of knowledge (Xing et al., 2003) to counteract this narrow view.