Integrating Rules and Connectionism for Robust Commonsense Reasoning

One of the more difficult problems for artificial intelligence research involves modelling commonsense reasoning. Traditional models have great difficulties in capturing the flexible and robust nature of commonsense reasoning. This book tackles the problem by adopting innovative approaches that use models consisting of a network of simple, interconnected elements, called neural networks. Using such models, rule-based algorithm reasoning and similarity-based comparison reasoning can be integrated.