基于三元组特征和词向量技术的中文专利侵权检测研究*

无论是在专利申请前还是在侵权诉讼中,专利侵权检测都能起到重要的作用,帮助企业或个人有效规避侵权和第三方侵权的风险。针对中文专利侵权检测中关键词特征表达能力弱以及句子结构特征容易引起噪声干扰的问题,提出了一种通过抽取三元组特征来改进中文专利侵权检测的方法,该方法将专利权利要求书抽取为三元组特征的集合,并结合词向量和HowNet计算三元组特征间的语义相似度,从而有效提高对疑似侵权专利的识别能力。实验结果表明,该方法取得了较好的检测效果,且在准确率上要高于其他方法。

Infringement detection of Chinese patent based on three tuple character and word embedding

Whether it is beforeSapplyingSforSaSpatent or in tort action, patent infringement detection can play an important role, so as to help enterprises or individuals to effectively circumvent the risk of infringement and the third party infringement.Because the expression ability of keywords features are weak and the structural features of the sentence are easy to cause the problem of noise interference,this paper proposed a method of improving Chinese patent infringement detection by extracting the three tuple features of the claim.In this method,the patent claim was extracted a set of three tuple, and the similarity between the three tuple features was calculated by combining word embedding and HowNet, which can effectively improve the ability to identity the suspected patent infringement. Experimental results show that the proposed method has good detection results ,and the accuracy is higher than other methods.