CAJ | 학술논문

知识产权作为一种重要的生产要素，得到广泛关注，专利侵权检测是规避侵权和防范第三方侵权的有效手段。但专利信息具有规模大、专业程度高的特点，增加了侵权检测的难度。专利侵权检测通常以基于专利文本向量的相似度测量为主，但本文考虑到大量专利数据的环境下，中文专利文本向量呈现出高维稀疏的特性，难以通过相似度的测量达到高效检测的目标，因此，本文提出采用基于SOM算法的模糊聚类对专利侵权进行判定，为专利审查人员和专利的利益相关者提供可视化程度高、技术特点细分的专利侵权检测方法。
지식산권작위일충중요적생산요소，득도엄범관주，전리침권검측시규피침권화방범제삼방침권적유효수단。단전리신식구유규모대、전업정도고적특점，증가료침권검측적난도。전리침권검측통상이기우전리문본향량적상사도측량위주，단본문고필도대량전리수거적배경하，중문전리문본향량정현출고유희소적특성，난이통과상사도적측량체도고효검측적목표，인차，본문제출채용기우SOM산법적모호취류대전리침권진행판정，위전리심사인원화전리적이익상관자제공가시화정도고、기술특점세분적전리침권검측방법。
Intellectual property right (IPR) is widely considered as a vital productive factor. Patent infringement detection is an effective approach to avoid infringement and strengthen the precaution of third-party infringement. Patent documents distribute in a large scale and cover a wide range of professional information;therefore, there exists difficulty in patent infringement detection. When solving the issue a-bove, historical study often focuses on similarity measurement on patent document vectors. But considering the matrix's feature of high-di-mensional and sparse in the condition of big data, it is difficult to improve the efficiency of infringement detection. In order to provide pa-tent infringement detection method with remarkable visualization capability and subdivision of technical features for patent examiners and stakeholders concerned with IPRs, we propose a fuzzy cluster method of patent infringement judgment based on SOM.