摘要
针对微博舆情信息的特点,提出基于语义理解的微博舆情信息关联检测方法。从舆情信息表示模型和舆情信息相关度计算方法两个方面展开研究。在信息表示方面,使用微博的评论信息扩充微博信息以期较好地应对数据稀疏现象,基于同义词词林来计算词汇相似度,以应对微博草根性带来的问题,将微博舆情信息表示成多个向量空间模型。在相关性计算方面,提出多维度相关性计算方法。实验证明,所提出的方法对关联检测的准确率和召回率都有较好的提升。
Based on the features of micro-blog public opinion, this paper proposed a micro-blog public opinion link detection method based on semantic mining. This method focused on the representation model and the similarity computing method of public opinion. In terms of the representation model, responses to a micro-blog post were used to expand this micro-blog post with the aim to deal with the problem of data sparseness, and then based on the Tongyici Cilin,the similarity between words was computed with the aim to deal with the problem of micro-blog's grassroot nature,Finally,the multi-vector space model was established to represent the public opinion. In terms of the similarity computation method, a multi-dimensional similarity computation method was proposed, gxperirnental results show that our proposed method can improve the precision and recall of the micro-blog public opinion link detection effectively.
出处
《山东科技大学学报(自然科学版)》
CAS
2015年第4期62-66,共5页
Journal of Shandong University of Science and Technology(Natural Science)
基金
青岛市科技计划项目(12-1-4-6-(9)-jch)
关键词
微博
语义
关联检测
微博舆情
多向量空间模型
草根
microblog
semantic mining
link detection, micro blog public opinion
multivector space model
grassroot