机械产品专利作用结构知识提取方法研究

Acquisition Method for Principle Solution of Mechanical Patent

摘要:

机械产品专利包含了产品改进和创新的技术知识，对专利知识有效提取和合理表示是利用专利知识辅助产品创新的关键技术之一。分析了现有专利技术方案提取方法和知识表示所存在的问题，提出了针对英文专利包括技术对象和技术关系两方面提取任务的专利作用结构知识提取方法，建立了基于可扩展标记语言结构XSD表示和统一建模语言UML的专利作用结构知识表示模型。研究了机械产品专利语言的特点，采用最大熵原理和专利术语词典识别提取技术对象，通过建立组成类动词库识别核心动词提取技术关系。该专利作用结构知识表示模型具有表达直观和提取过程计算机自动处理等特点。结合专利实例，提出了从专利中提取专利作用结构知识，写入XSD结构化文档，最后转化为专利作用结构表示图的获取过程。

Abstract:

The problems of the existed acquisition method and representation method for patent technical solution were discussed. A new knowledge acquisition method for mechanical patent’s principle solution including the acquisition task of technical object and technical relationship was proposed. Based on extensible markup language schema definition (XSD) and unified modeling language (UML), the representation model of patent technical solution was built up. According to the characteristics of mechanical patent’s text, technical object with the method of maximum entropy principle and patent term dictionary was acquired. The composition verb database was established and technical relationship was acquired. This method has some advantages such as the intuitionistic representation of the patent knowledge and the feature that the knowledge was automatically processed by computer. A whole process involved the steps like acquiring the principle knowledge from patent documents, writing the knowledge into XSD and transforming XSD to the representation model was illustrated with a US patent in detail. The model and process laid the foundation for the efficient utilization of patent knowledge.