The rest of this thesis is organized as follows. Chapter 2 gives a description of the algorithm employed by the BruteDL system. Chapter 3 describes the experiments on BruteDL and the performance of its default strategy. A detailed description and experimental results of three alternative default strategies are given in Chapters 4 through 7. Chapter 4 presents the Minimum OSR strategy. Chapter 5 and 6 deal with the method of storing all the homogeneous rules regardless of whether they are the best for at least one training example or not. Chapter 7 describes the method of using the closest matching instead of using the default rule. Finally, in Chapter 8, conclusions drawn from our experiments are discussed.