A large number of Web information extraction algorithms are based on machine learning techniques. For such extraction algorithms, we propose employing a lazy learning strategy to build a specialized model for each test ...