基于情感Ontology的资源分析模型  

Resource Analysis Model Based on Sentiment Ontology

作　　者：刘倩[1] 陶县俊[1] 王晓东[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,新乡453007

出　　处：《计算机与数字工程》2009年第9期115-119,共5页Computer & Digital Engineering

基　　金：河南省科技攻关计划项目(编号:082102210007)资助

摘　　要：对资源分析方法进行了研究,并提出了一种基于情感Ontology的分析方法。首先基于"知网"构建情感On-tology,然后基于情感Ontology抽取资源分析的特征词汇并判断其情感倾向性,最后根据抽取的特征词汇对整篇文本的情感倾向进行分析。实验结果表明,在以人工标注做Baseline的基础上,利用情感Ontology抽取特征词汇的资源分析方法可以使情感识别的准确率达到78.87%。Methods of resource analysis were studied and present an approach based on Sentiment Ontology, First Sentiment Ontology is constructed with the sentiment words that HowNet has tagged, and then select the features from an undetermined document and estimate their orientation, at last analysis the orientation of the entire document. The experimental results show that the sentiment identification accuracy reached to 78.87% using the features extracted according to Sentiment Ontology compared with the Baseline that identifying the sentiment orientation all by manual work.

关 键 词：ONTOLOGY 文本倾向性分析 “知网” 词汇相似度 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]