中文摘要：

模式匹配是模式集成、语义WEB及电子商务等领域的重点及难点问题．为了有效利用专家知识提高匹配质量，提出了一种基于部分已验证匹配关系的模式匹配模型．在该模型中，首先，人工验证待匹配模式元素问的少量对应关系，进而推理出当前任务下部分已知的匹配关系及单独匹配器的缺省权重：然后，基于上述已收集到的先验知识对多种匹配器所生成的相似度矩阵进行合并及调整，并在全局范围内进行优化：最后，对优化矩阵的选择性进行评估，从而为不同匹配任务推荐最合理的候选匹配生成方案．实验结果表明，部分已验证匹配关系的使用有助于模式匹配质量的提高．

英文摘要：

Schema matching is an important and difficult problem in many database application domains, such as data integration, semantic web and data warehousing and so on. In order to use expertst knowledge to improve the matching quality effectively, a schema matching model is proposed based on partial verified matching relations. In this model, first, a small amount of correspondences between schemas elements are verified by manual, and the partial known matching relations and default weights of different matchers are reasoned on the current task; second, the similarity matrices of multiple marchers are combined based on the collected priori knowledge, and optimized under global scope; finally, the selectivity of the optimization matrix is evaluated, and the most reasonable candidate matching generation plans for different matching tasks are generated. Experimental results show that the use of partial verified matching helps to improve the quality of schema matching.