基于无指导学习的微博评论分析方法

发布时间：2018-08-28 07:46

【摘要】：该文以一种有效的方法寻找出有价值的微博评论,这对于读者更高效地阅读评论,为舆情分析、文本挖掘等任务提供支持,均具有重要的应用价值。针对微博及其评论文本短小、内容发散等特点,该文提出一种基于无指导学习的微博评论分析方法,该方法通过互联网搜索引擎扩展微博文本,基于相关性计算自动构造正负训练用例,生成特定的某条微博评论分类模型,通过该模型对评论的价值性进行评估。实验结果表明,该方法能够比较好地识别出评论的价值。
[Abstract]:This paper uses an effective method to find valuable Weibo comments, which has important application value for readers to read comments more efficiently, to provide support for public opinion analysis, text mining and other tasks. In view of the characteristics of Weibo and his comments, such as short text and divergent content, this paper proposes an analysis method of Weibo's comments based on unguided learning. This method extends Weibo's text through the Internet search engine. The positive and negative training cases are automatically constructed based on the correlation calculation, and a specific Weibo comment classification model is generated, through which the value of comments is evaluated. The experimental results show that the method can recognize the value of comments.
【作者单位】：南京大学计算机软件新技术国家重点实验室;
【基金】：国家自然科学基金(61170181) 江苏省自然科学基金(BK2011192) 国家社会科学基金(11AZD121)
【分类号】：TP391.1

【相似文献】