中文专利文献术语抽取  

Chinese patent terminology extraction

作　　者：徐川[1] 施水才[1,2] 房祥[1,2] 吕学强[1,2] 

机构地区：[1]北京信息科技大学网络文化与数字传播北京市重点实验室,北京100101 [2]北京拓尔思信息技术股份有限公司,北京100101

出　　处：《计算机工程与设计》2013年第6期2175-2179,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(61171159);国家科技支撑计划课题基金项目(2011BAH11B03);北京市教委科技发展计划基金项目(KM201110772021;KM201211232023)

摘　　要：为了有效解决专利文献中术语抽取问题,提出采用字符串之间的结合强度融合词性过滤法抽取术语的方法。根据专利文献中术语出现的特点,提出了字符串之间的边界结合度方法和字符串之间的串边结合度方法,度量了字符串间的结合强度。在此基础之上,根据术语内部结构中词性的组成特点,提出了术语双字词性过滤的方法,并与结合强度的计算方法进行融合。实验结果表明,该方法对中文专利术语抽取有很好的效果,平均正确率为80.24%,平均召回率为80.61%。To deal with the problem of patent terminology extraction, a method based on the bonding strength and character filtering is proposed. Based on characteristics of patent terms, the combination degree of boundary-to-boundary and string-to-boundary of strings are proposed to measure the bonding strength. Based on these, a method for filtering two character terms is pro-posed according to the internal structure of part-of-speech in the term, and is combined with the calculating method which is based on bonding strength. Experiments show that the proposed method is practicable on extracting Chinese patent terms. The average precision is 80.24% and the average recall is 80.61%.

关 键 词：术语 边界结合度 串边结合度 双字词性过滤法 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]