文章摘要
张海超,赵良伟.利用Doc2Vec 判断中文专利相似性[J].情报工程,2018,4(2):064-072
利用Doc2Vec 判断中文专利相似性
Judge Chinese Patents Similarity Based on Doc2Vec
  
DOI:10.3772/j.issn.2095-915X.2018.02.007
中文关键词: 专利相似度  专利侵权  Word2Vec  Doc2Vec
英文关键词: Similarity of Patents  patents Infringement  Word2Vec  Doc2Vec
基金项目:基于文本内容相似性的中文专利侵权判定方法研究(YY2016-04)
作者单位
张海超 中国科学技术信息研究所国家科技信息资源综合利用与公共服务中心 
赵良伟 邢台职业技术学院 
摘要点击次数: 263
全文下载次数: 276
中文摘要:
      目前专利侵权纠纷案件时有发生,企业一旦卷入专利侵权纠纷,通常会面临时间考验和经济损失。本文选取中文专利数据样本,抽取专利权利要求书形成训练语料,并利用Doc2Vec 深度神经网络算法,计算权利要求书文本之间的相似度,得出与涉案专利相似性较高的专利。并且将上述方法应用到专利复审案件实验中,进行实证研究,取得了较好的效果。需要进一步提高训练数据的质量,对比其他算法的效果。利用该方法能够帮助专利审查人员和企业找到相似专利。
英文摘要:
      Recently, patent infringement disputes occurred frequently. Once a company was involved in a patent infringement dispute, it usually faced the time test and economic loss. This paper chose the Chinese patents as data source, extracted the patent claims as the training corpus, and used the Doc2Vec deep neural network algorithm to calculate the similarity between the claims. Then, we obatined the patent with higher similarity to the involved patent. Finally, the above methods was applied to the patent reexamination case to conduct the empirical research. The results indicated that this method can achieve good results.Moreover, the results also suggested that the method need to be further improved based on the high quality of training data and comparison with other algorithms. This method may help the patent reviewers and enterprises to find the similar patents.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮