Publisher Summary The correct specification of knowledge is supreme for the success of any autonomous system. The most common type of intelligent system that can be defined by using symbolic processing (SP) tools is an expert system (ES), which is a knowledge-based program that provides expert quality solutions to problems in a specific domain. The knowledge manipulated by the system is incorporated into its main component, the knowledge base, which is built using one or more of the existing knowledge representation languages. Each of those representation schemes has its own particularities, and the variety of forms each can take makes it necessary to categorize the type of problems requiring an ES, so that the most appropriate knowledge representation form(s) (KRF) is used. Designing an autonomous intelligent system that incorporates a knowledge base built with the tools whose underlying theory embraces the connectionist paradigm requires the use of at least one of the various neural network (NN) models defined to perform neural computation. This chapter reviews existence of important differences between the representation forms derived from paradigms, and demonstrates the various useful results obtained through their integration. It outlines the KRFs considered relevant in this study. It also provides the advantages and/or disadvantages that result from the process of reproducing the properties of the KRFs.