To make up the limitations of existing schema matching methods based on schema structure information, a schema matching model called SKM (schema and reused knowledge based matching model) is proposed based on schema structure information and known matching knowledge. In this model, neural network influence procedure is imitated to realize semantic matching reasoning. The known matching knowledge is reused to mine the deep semantic relation between two schemas. It is also reused to curtail uncertain threshold interval automatically to specify the threshold for decreasing manual intervention. A simple approach of specifying matching relation between two matching elements is given. In the meantime, a self-learning adaptive and iterative model is presented to mine and enrich the known matching knowledge. Experimental results show that the SKM is feasible.