191_基于多源知识和ranking svm的中文微博命名实体链接.pdfVIP

文章编号：1003-0077 （2011）00-0000-00 基于多源知识和Ranking SVM 的中文微博命名实体链接 陈万礼，昝红英，吴泳钢 （郑州大学 信息工程学院，河南 郑州 450001 ） 摘要：命名实体在文本中是承载信息的重要单元，正确分析存在歧义的命名实体，对文本的理解起到关键 性作用。本文提出基于多源知识和Ranking SVM 的中文微博命名实体链接，结合同义词词典、百科资源等 知识产生初始候选实体集合，同时从文本中抽取多种组合特征，利用Ranking SVM 对候选实体集合进行排 1 序，从而得到目标实体。在 NLPCC2014 中文微博实体链接评测数据集上进行了实验，获得了89.40%的微 平均准确率，与NLPCC2014 中文微博实体链接评测取得最好成绩的系统相比，本文的系统具有一定的优势。 关键词：命名实体；中文微博实体链接；同义词词典；百科资源；Ranking SVM ；语义特征 中图分类号：TP391 文献标识码：A Chinese Micro-blog Named Entity Linking Based on Multisource Knowledge and Ranking SVM algorithm CHEN Wanli, ZAN Hongying,WU Yonggang (School of Information Engineering, Zhengzhou University, Zhengzhou Henan, 450001, China) Abstract: Named entity is an important component conveying information in texts; so an accurate understanding of the named entities is needed to ensure a correct analysis of the text information. This paper proposes a Chinese micro-blog entity linking strategy based on multi-resource knowledge and Ranking SVM algorithm, combing a dictionary of synonyms, the encyclopedia resources to produce an initial set of candidate entities , then various combinations of features extracted from the text and the use of Ranking SVM algorithm to generate the second sort of candidate entity set.In this strategy, named entities to be linked in micro-blog are mapped to the corresponding candidate entities in the knowledge base. The evaluation results gain a micro average accuracy of 89.04%, based on experiments using data sets of NLPCC2014 Chinese micro-blog entity linking track. Compared with the state-of-the-art result, the accuracy of this method demonstrates the effectiveness of our method. Keywords: named entity; chinese micro-blog entity linking; dictionary of synonyms; encyclopedia resources; Ranking SVM; semantic features 1 引言