摘要
在对语料库中表达产品特征及相应评价的词的词性进行分析的基础上,确定了表达产品特征及评价最为常见的词性和词性的重要程度顺序,提出了一种产品特征及其相应评价的信息抽取规则,并根据规则建立评价语句的语义倾向的计算公式。实验结果表明,该方法在产品特征抽取及其相应评价的语义倾向判断上具有很高的准确性。通过对产品特征及其相应的评价信息进行挖掘可以为企业新产品的开发和产品的推荐提供重要的参考价值,是进行下一步生产决策的重要的理论依据。
Based on analysis of part of speech of word which can express the character of product and the corresponding review in corpus, the most frequent part of speech and the corresponding order of importance is determinted, a new information extraction rules of the character of product and the corresponding review is proposed, and the formula of computering of semantic of sen- tence is established according to the rules. Experiment show that this method have a high accuracy in extracting of the character of product and computing of semantic orientation of the corresponding of review. It will provide a huge value of new product's development and product recommendation in enterprise and as a important theoretical for the next step of product decision.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第1期283-288,共6页
Computer Engineering and Design
基金
北京市哲学社会科学"十一五"规划基金项目(10AbJG389)
关键词
抽取规则
分词
语义倾向
同义词替换
信息挖掘
extraction rules
segment
semantic orientation
synonymous substitutiom information mining