基于文本的关键词提取方法研究与实现  

Research and Implementation of Text-based Keyword Extraction

作　　者：姜永清 赵宪佳 Jiang Yongqing;Zhao Xianjia(School of Data Science and Software Engineering,Qingdao University,Qingdao Shandong 266071,China)

机构地区：[1]青岛大学数据科学与软件工程学院,山东青岛266071

出　　处：《信息与电脑》2020年第5期51-54,共4页Information & Computer

摘　　要：关键词提取技术是目前计算机文本分类等技术采用的基本技术。笔者根据传统TFIEF算法的特点,研究出创新型的关键词提取算法,算法的基础是以词语权重来计算的,并在文本候选关键词中把词频比设计成去噪音的权值以及引入词语相似度,该方法对降低文本同类语料库提取关键词权重的影响都是非常有效的,也消除了TFIDF算法存在偏差。经过实验的分析和对比,该新型方法拥有更高的平均召回率。Keyword extraction is a basic technique used in computer text classification.According to the characteristics of some traditional TFIEF keyword extraction algorithm,I have developed innovative keyword extraction algorithm,this keyword extraction algorithm based on weight to calculate,and put the word in the text the candidate keywords than designed to noise of weights and introduced words known degrees,such an approach to reduce the effects of text similar weights corpus to extract the keywords are very effective,also eliminates the deviation of flawed TFIDF algorithm.After experimental analysis and comparison,this new method does have a higher average recall rate.

关 键 词：TFIDF算法 关键词提取 词语相似度 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]