In an XML database, finding all occurrences of a twig pattern is a core operation for XML query processing. In the past few years, many algorithms, such as Holistic Twig and TJFast, were proposed in theliteratures. However, these algorithms are based on merging, with high computational cost. Recently Twig2Stackalgorithm and TwigList algorithm are proposed to resolve this problem, but they are very complex. Aim at this problem, this paper considers the characteristic that most path expressions have only a few output nodes, and proposes two new algorithms without merging, named TwigNM and TwigNME, which use only a few stacks.Finally, the experimental results show that these algorithms are superior to the previous algorithms, especially for only ancestor-descendant relationship in XPath.