专利技术功效短语获取研究

在专利技术功效矩阵构建研究中,专利技术功效短语获取是矩阵构建的基础,也是构建矩阵的词汇来源。专利技术功效短语获取的准确性直接影响专利技术功效矩阵构建的效果。为了提高专利技术功效短语的准确性,基于汽车新能源专利文献文本数据基础上,综合考虑专利文献结构、专利文献线索词,以及专利文献的句法、语法分析等多种因素,提出了基于规则和统计相结合的专利技术功效短语获取方法。首先,根据专利摘要文本定位包含专利技术功效短语的单句,提取技术功效目标句;其次,在改进的分词方法和词性标注的基础上,针对包含功效短语的句子,结合依存关系规则、短语规则计算出共现频率较高的词,并提取技术功效短语。利用该方法获取专利技术功效短语,其准确率可到达85%。实验证明该方法在获取专利技术功效短语中是有效的、可行的,进而整体上提高专利技术功效短语的识别效果。

Extracting technologies efficacy phrases of patent for research

In the study of building a patent technology effect matrix, gaining patent technology effect phrase is the basis of the matrix to construct, also is the word source of building matrix. Gaining patent technology effect phrase directly affects the accuracy of patent technology efficiency matrix to construct effect. In order to improve the accuracy of the patent technology efficacy phrases, based on the new energy automobile patent literature text data, based on the comprehensive consideration of patent document structure, patent document clues, and syntax and grammar analysis of patent literature and other factors, is proposed based on rules and statistics with the combination of patent technology effect phrase gain method. First of all, according to the patent contains the text location patent technology efficacy phrase sentence, extraction technology efficacy target words; Second, the improvement on the basis of word segmentation and part-of-speech tagging, aimed at containing efficacy phrase sentence, combination rules, phrase, calculate the interdependence between the co-occurrence frequency word, efficacy phrases and extracting technology. This article USES the method to obtain patent technology efficacy phrases, its accuracy can reach 85%. Experiment proved that the method in obtaining the patent technology efficacy phrase is effective and feasible, thus increase the effectiveness of patent technology on whole phrases recognition effect.