摘要
互联网中包含着海量的用户评论,这些评论中包含着用户对产品、服务、品牌、厂商的观点和情感倾向等主观信息,是一种重要的竞争情报。但由于自然语言中人类情感表述的复杂和微妙,现有的竞争情报系统难以实现对这些主观信息的挖掘和利用。在现有情感分析和观点挖掘研究的基础上,本文提出了一种观点挖掘方法,利用手工标注的语料库和针对特定领域的知识树,提高了现有观点挖掘方法的性能。并利用这种方法建立了一个竞争情报系统,自动采集互联网中海量的用户评论信息,从中挖掘用户观点,探讨了对挖掘结果进行分析的方法,并展示了分析结果可视化生成的情报产品。
In the Internet there is huge amount of user reviews, which contain subjective user opinions and can be taken as an important competitive intelligence source. However, because of the complicated and delicate expression of human sentiment in natural language, it' s hard for current CIS to utilize the user reviews. Based on the current research of sentiment analysis and opinion mining, an opinion mining method is proposed in this paper. The method improves the performance of opinion mining with a manual annotated corpus and a knowledge tree of the domain which the extracted documents belong to. Based on the opinion mining method proposed, a competitive intelligence system is established, which automatically collects huge amount user reviews from the Internet, and extracts user opinions from the documents collected. The approaches of intelligence analysis based on the opinion data extracted are also discussed, and the intelligence products generated by visualization of the analyzing resuh are demonstrated.
出处
《情报学报》
CSSCI
北大核心
2012年第2期174-179,共6页
Journal of the China Society for Scientific and Technical Information
基金
本文受国家自然科学基金项目“基于互联网网民言论信息的口碑监测、分析与管理研究”(71073006)支持.