Because web and wireless mobile users of geographical information systems require information that is directly relevant to the specific task in which they are engaged, Traditional generalization techniques produce maps that are general-purpose presents a new set of challenges with the growth of web and wireless of geographical information systems. Therefore, a possible solution to the problem relies on applying task-oriented generalization techniques including simplification adaptive to the diverse requirements for users of web or wireless mobile GIS. In this paper, the level of detail in the generated web mapping is adapted by progressive generalization algorithm, which remains or reduces online web map description data according to the geometric and semantic of the features represented. At the same time, we research topological checking method for web mapping to avoid topological conflicts. Based on this, we generate web mapping on demand and prove effective generalization methods by concrete experiments.