CAJ | 학술논문

随着专利数据规模的不断增长,对专利数据的深入挖掘也变得日益重要,特别是专利数据中所蕴含的技术功效等信息具有较高的价值.本文提出了一种基于隐马尔科夫模型的专利功效词识别方法,通过词法与句法分析筛选出候选功效词,在此基础上,采用隐马尔科夫模型并结合专利发明改进的特征设计了功效词识别算法,对候选功效词进行过滤.在新能源汽车等不同领域的专利数据集上,以准确率与召回率作为评价标准,验证所提出方法的有效性.实验结果表明,此方法有效提高了识别准确率与召回率.
수착전리수거규모적불단증장,대전리수거적심입알굴야변득일익중요,특별시전리수거중소온함적기술공효등신식구유교고적개치.본문제출료일충기우은마이과부모형적전리공효사식별방법,통과사법여구법분석사선출후선공효사,재차기출상,채용은마이과부모형병결합전리발명개진적특정설계료공효사식별산법,대후선공효사진행과려.재신능원기차등불동영역적전리수거집상,이준학솔여소회솔작위평개표준,험증소제출방법적유효성.실험결과표명,차방법유효제고료식별준학솔여소회솔.
With the development of the patent data, the technique of patent data mining becomes more important, especially the technical efifciency information entailed in the patent data which have the higher value. We put forward the method to recognize the efifcacy word based on the hidden Markov model. We select the candidate efifcacy word by the use of lexical and syntactic analysis approach ifrstly. The recognition algorithm is designed by the combination of Hidden Markov model and features in the patent data. We give the experiment in different patent ifelds and the metric of precision and recall are used for the evaluation. The experimental result shows that our method gets the better performance.