基于组合词和同义词集的关键词提取算法  

Keywords extraction algorithm based on combined word and synset

作　　者：蒋昌金[1] 彭宏[1] 陈建超[1] 马千里[1] 严桂夺[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广州510640

出　　处：《计算机应用研究》2010年第8期2853-2856,共4页Application Research of Computers

基　　金：广东省自然科学基金资助项目(07006474;9451064101003233);广东省科技(2007B010200044)

摘　　要：为了提高关键词的提取准确率,在对现有关键词抽取方法进行研究的基础之上,针对影响关键词提取准确率的分词技术、同义词现象等难点,提出了一种基于组合词和同义词集的关键词提取算法。该算法首先利用组合词识别算法极大地改进分词效果,能识别网页上绝大多数的新词、未登录词,为提高关键词自动抽取准确率奠定了坚实的基础;同时利用构造的同义词集,合并同义词的词频,避免了同义词在输出结果中同现;利用综合评分公式,充分考虑候选关键词的位置、长度、词性等特性。实验数据表明,该方法有较高的提取准确率。This paper presented a Chinese webpage keywords extraction algorithm after the study of existing techniques for keyword extraction. The presented approach could extremely improve the performance of Chinese word segmentation system. The modified Chinese word segmentation system could recognise most of new terms,phrases and non-login words in Chinese webpage and this is vitally important for Chinese keyword extraction. Moreover,constructed a synset database and used adding the frequencies of synonyms together,avoiding the co-occurrence of synonyms in output. Further more,created a eva-luation function to score candidate keyword based on its location,length,part-of-speech. The experiment results show that the proposed algorithm has better performance compared with the traditional keyword extraction algorithms.

关 键 词：组合词 同义词集 中文网页 关键词提取 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]