中文摘要：

针对中国国家知识产权局专利审查中的专利检索任务,设计了一个两步专利检索模型：第一步进行布尔初步检索;第二步采用向量空间模型进行相似性计算,从而得到排序的二次检索结果。实验数据来源于中国国家知识产权局,结果显示该检索模型是有效的,部分地克服了现有检索方法的不足,大大减轻了审查员的工作负担,提高了工作效率。

英文摘要：

The paper proposed a two-step Chinese patent search model for examination task in State Intellectual Property Office of China. At 1-step Boolean query was used to improve recall, at 2-step used vector space model to sort the results retrieved at 1-step by computing similarity between applying patent（query） and previous patents（documents）. Experiments data set comes from SIPO. Evaluation results show that the model works well; it overcomes drawbacks of existing patent search method for examination task partially, improves the examiner＇ s efficiency.