中文摘要：

[目的／意义]在专利分析中引入Knowledge Graph，将专利内容转换为由Knowledge Graph中实体语义关系所构成的图结构，进而探索该形式的专利表示方法在识别专利诉讼案中专利证据的可行性。[方法／过程]在专利内容转换过程中，首先采用自动术语识别方法提取其实体指称，并通过实体链接将实体指称转化为命名实体，进而根据图算法识别出该专利的隐含实体，最终形成该专利所对应的图结构。[结果／结论]将该专利表示方式应用于硬盘驱动器领域来寻找专利诉讼案中可用的证据专利，实证结果表明，与当前主流的专利文本表示方式相比，该方法在寻找证据专利效果上有较大提升。

英文摘要：

[ Purpose/significance] This paper introduces knowledge graph to patent analysis, and it transforms pa- tent content from unstructured text to graph structure with node as entity and edge as semantic relationship. Furthermore, the feasibility of patent evidence recognition is explored. [ Method/process] During transformation of patent content, we use ATE （ Automatic Terminology Extraction） method to find entity mentions from patent text, and change them into entity via entity linking based on knowledge graph. Then we use the proposed graph algorithm to recognize hidden entities in pa- tent, and finally output the patent＇ s graph structure. [ Result/conclusion ] We apply this presentation to hard disk drive to find potential patent evidence, and empirical result shows that the proposed presentation method of patent content can outperform current mainstream method to a large extent.