基于SAO的中文相似专利识别方法及其应用研究

发布时间：2019-02-16 06:54

【摘要】：现如今,专利侵权纠纷案件时有发生。企业一旦卷入其中将面临着巨额经济损失。面对专利侵权纠纷,企业最有效的措施是对涉案专利向专利复审委员会提起涉案专利无效请求。专利无效的主要证据就是在涉案专利申请日之前就已经有相似专利。因此,本文的研究意义在于利用提出的基于SAO的中文相似专利识别算法帮助企业找到相似专利,作为专利无效的证据,使之能够在专利侵权纠纷中占据主动。权利要求书是专利文献的核心内容,专利权的保护范围以权利要求的内容为准,专利侵权判定也是以权利要求书为依据。本文深入研究中文专利权利要求书的文本结构特点,提出一种中文相似专利识别算法,帮助涉案企业的专利无效请求。论文主要内容包括:中文专利权利要求书信息抽取。该部分首先将专利权利要求书进行分类、分割子串和结构化,然后对其进行分词、词性标注、句法分析和语义分析等一系列的处理。中文专利权利要求SAO结构的抽取与专利相似性计算。进行信息抽取后,从涉案专利和相关专利的权利要求书中抽取出SAO结构,计算SAO结构的相似度,由SAO结构的相似度得出专利文档相似度,并且转化成相异矩阵,进行多维尺度分析(MDS),判断专利间的相似性。实证分析。为了验证论文方法的有效性,选取了在地铁疏散平台方面的专利侵权诉讼实际案例。利用本文提出的中文相似专利识别算法,找到了企业提出专利无效请求的证据专利,更重要的是找到了同样可以作为无效证据的其他相似专利,验证了本文方法的有效性。
[Abstract]:Nowadays, patent infringement disputes occur from time to time. Once the enterprise gets involved in it, it will face huge economic losses. In the face of patent infringement dispute, the most effective measure for the enterprise is to file a patent invalidation request to the patent review board. The main evidence of invalidity is that similar patents were available before the date of filing of the patent in question. Therefore, the significance of this study is to use the proposed Chinese similar patent recognition algorithm based on SAO to help enterprises find similar patents, as evidence of patent invalidity, so that they can take the initiative in patent infringement disputes. Claim for rights is the core content of patent documents, the scope of patent protection is based on the content of claim, and the judgment of patent infringement is based on claim. In this paper, the text structure of the Chinese patent claims is deeply studied, and a Chinese similar patent recognition algorithm is proposed to help the enterprises concerned with invalid patent requests. The main content of the thesis includes: Chinese patent claim information extraction. In this part, the patent claims are classified, divided into substrings and structured, then a series of processing such as participle, part of speech tagging, syntactic analysis and semantic analysis are carried out. Chinese patent claim SAO structure extraction and patent similarity calculation. After information extraction, the SAO structure is extracted from the patent involved and the relevant patent claims, the similarity of the SAO structure is calculated, and the similarity of the patent document is obtained from the similarity of the SAO structure and transformed into a different matrix. A multidimensional scale analysis of the similarity between (MDS), judgment patents is carried out. Empirical analysis. In order to verify the effectiveness of the method, the actual case of patent infringement litigation in subway evacuation platform is selected. By using the Chinese similar patent recognition algorithm proposed in this paper, we find the evidence patent for the enterprise to put forward the patent invalidation request, and more importantly, find other similar patents which can also be used as invalid evidence, which verifies the validity of the method in this paper.
【学位授予单位】：北京工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：D923;D923.42

【相似文献】